| {: , : 2016, : , : , : [], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2016, : , : , : [, ], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2016, : , : , : [], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2016, : , : , : [, , onMichel\n Verhaegenabstract Linear Quadratic Gaussian (LQG) systems are well-understood and methods to\nminimize the expected cost are readily available. Less is known about the\nstatistical properties of the resulting cost function. The contribution of this\npaper is a set of analytic expressions for the mean and variance of the LQG\ncost function. These expressions are derived using two different methods, one\nusing solutions to Lyapunov equations and the other using only matrix\nexponentials. Both the discounted and the non-discounted cost function are\nconsidered, as well as the finite-time and the infinite-time cost function. The\nderived expressions are successfully applied to an example system to reduce the\nprobability of the cost exceeding a given threshold.\ndoi10.1016/j.automatica.2016.01.030arxiv_id1602.02524pdf_urlshttps://arxiv.org/pdf/1602.02524githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 67, May 2016, Pages 216-223categoriescs.SY |
| titleNew Stability and Exact Observability Conditions for Semilinear Wave\n Equationsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsEmilia FridmanMaria Terushkinabstract The problem of estimating the initial state of 1-D wave equations with\nglobally Lipschitz nonlinearities from boundary measurements on a finite\ninterval was solved recently by using the sequence of forward and backward\nobservers, and deriving the upper bound for exact observability time in terms\nof Linear Matrix Inequalities (LMIs) [5]. In the present paper, we generalize\nthis result to n-D wave equations on a hypercube. This extension includes new\nLMI-based exponential stability conditions for n-D wave equations, as well as\nan upper bound on the minimum exact observability time in terms of LMIs. For\n1-D wave equations with locally Lipschitz nonlinearities, we find an estimate\non the region of initial conditions that are guaranteed to be uniquely\nrecovered from the measurements. The efficiency of the results is illustrated\nby numerical examples.\ndoi10.1016/j.automatica.2015.10.008arxiv_id1602.07567pdf_urlshttps://arxiv.org/pdf/1602.07567githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 63, pp. 1-10, 2016categoriescs.SY math.AP |
| titleA jammer's perspective of reachability and LQ optimal control", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Sukumar Srikant and Debasish Chatterjee"], "abstract": " This article treats two problems dealing with control of linear systems in\nthe presence of a jammer that can sporadically turn off the control signal. The\nfirst problem treats the standard reachability problem, and the second treats\nthe standard linear quadratic regulator problem under the above class of\njamming signals. We provide necessary and sufficient conditions for optimality\nbased on a nonsmooth Pontryagin maximum principle.\n", "doi": "10.1016/j.automatica.2016.03.026", "arxiv_id": "1603.00573", "pdf_urls": ["https://arxiv.org/pdf/1603.00573"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Vol. 70, pp. 295-302, 2016", "categories": "cs.SY math.OC"} |
| {"title": "A flexible state space model for learning nonlinear dynamical systems", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Andreas Svensson and Thomas B. Sch\\\"on"], "abstract": " We consider a nonlinear state-space model with the state transition and\nobservation functions expressed as basis function expansions. The coefficients\nin the basis function expansions are learned from data. Using a connection to\nGaussian processes we also develop priors on the coefficients, for tuning the\nmodel flexibility and to prevent overfitting to data, akin to a Gaussian\nprocess state-space model. The priors can alternatively be seen as a\nregularization, and helps the model in generalizing the data without\nsacrificing the richness offered by the basis function expansion. To learn the\ncoefficients and other unknown parameters efficiently, we tailor an algorithm\nusing state-of-the-art sequential Monte Carlo methods, which comes with\ntheoretical guarantees on the learning. Our approach indicates promising\nresults when evaluated on a classical benchmark as well as real data.\n", "doi": null, "arxiv_id": "1603.05486", "pdf_urls": ["https://arxiv.org/pdf/1603.05486"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 80(2017), page 189-199", "categories": "stat.CO cs.SY stat.ML"} |
| {"title": "Moving horizon estimation for discrete-time linear systems with binary\n sensors: algorithms and stability results", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Giorgio Battistelli", "Luigi Chisci", "Stefano Gherardini"], "abstract": " The paper addresses state estimation for linear discrete-time systems with\nbinary (threshold) measurements. A Moving Horizon Estimation (MHE) approach is\nfollowed and different estimators, characterized by two different choices of\nthe cost function to be minimized and/or by the possible inclusion of\nconstraints, are proposed. Specifically, the cost function is either quadratic,\nwhen only the information pertaining to the threshold-crossing instants is\nexploited, or piece-wise quadratic, when all the available binary measurements\nare taken into account. Stability results are provided for the proposed MHE\nalgorithms in the presence of unknown but bounded disturbances and measurement\nnoises. Performance of the proposed techniques is also assessed by means of a\nsimulation example.\n", "doi": "10.1016/j.automatica.2017.07.035", "arxiv_id": "1604.00217", "pdf_urls": ["https://arxiv.org/pdf/1604.00217"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 85, 374-385 (2017)", "categories": "cs.SY"} |
| {"title": "Cascade and locally dissipative realizations of linear quantum systems\n for pure Gaussian state covariance assignment", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Shan Ma", "Matthew J. Woolley", "Ian R. Petersen and Naoki Yamamoto"], "abstract": " This paper presents two realizations of linear quantum systems for covariance\nassignment corresponding to pure Gaussian states. The first one is called a\ncascade realization; given any covariance matrix corresponding to a pure\nGaussian state, we can construct a cascaded quantum system generating that\nstate. The second one is called a locally dissipative realization; given a\ncovariance matrix corresponding to a pure Gaussian state, if it satisfies\ncertain conditions, we can construct a linear quantum system that has only\nlocal interactions with its environment and achieves the assigned covariance\nmatrix. Both realizations are illustrated by examples from quantum optics.\n", "doi": "10.1016/j.automatica.2017.12.061", "arxiv_id": "1604.03182", "pdf_urls": ["https://arxiv.org/pdf/1604.03182"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 90, 263-270, 2018", "categories": "quant-ph cs.SY math.OC"} |
| {"title": "Asymptotic behaviour in the robot rendezvous problem", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["L. Paunonen and D. Seifert"], "abstract": " We present a non-technical overview of the results obtained by the authors\n(2015) concerning the so-called robot rendezvous problem studied by Feintuch\nand Francis (2012). In particular, we present a necessary and sufficient\ncondition for convergence of the solution in terms of Ces\\`aro convergence of\nthe translates $S^k x_0$, $k\\ge0$, of the sequence $x_0$ of initial positions\nunder the right-shift operator $S$, thus shedding new light on questions left\nopen in by Feintuch and Francis. We also formulate a stronger ergodic condition\non $x_0$ which ensures that the corresponding solution converges to its limit\nat the rate $O(t^{-1/2})$ as $t\\to\\infty$. We conclude with a brief discussion\nof a natural two-sided variant of the robot rendezvous problem and by relating\nthe robot rendezvous problem to a more realistic model of vehicle platoons, for\nwhich the authors' earlier general results lead to analogous statements.\ndoi10.1016/j.automatica.2017.02.015arxiv_id1604.05553pdf_urlshttps://arxiv.org/pdf/1604.05553githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 79, 2017, pp. 127-130categoriesmath.OC math.FA |
| titleDistributed controller-estimator for target tracking of networked\n robotic systems under sampled interactionyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMing-Feng GeZhi-Hong GuanBin HuDing-Xin Heand Rui-Quan Liaoabstract This paper investigates the target tracking problem for networked robotic\nsystems (NRSs) under sampled interaction. The target is assumed to be\ntime-varying and described by a second-order oscillator. Two novel distributed\ncontroller-estimator algorithms (DCEA), which consist of both continuous and\ndiscontinuous signals, are presented. Based on the properties of small-value\nnorms and Lyapunov stability theory, the conditions on the interaction\ntopology, the sampling period, and the other control parameters are given such\nthat the practical stability of the tracking error is achieved and the\nstability region is regulated quantitatively. The advantages of the presented\nDCEA are illustrated by comparisons with each other and the existing\ncoordination algorithms. Simulation examples are given to demonstrate the\ntheoretical results.\ndoi10.1016/j.automatica.2016.03.008arxiv_id1605.08542pdf_urlshttps://arxiv.org/pdf/1605.08542githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2016, 69: 410-417categoriescs.SY cs.MA cs.RO |
| titleRecursive nonlinear-system identification using latent variablesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsPer MattssonDave Zachariah and Petre Stoicaabstract In this paper we develop a method for learning nonlinear systems with\nmultiple outputs and inputs. We begin by modelling the errors of a nominal\npredictor of the system using a latent variable framework. Then using the\nmaximum likelihood principle we derive a criterion for learning the model. The\nresulting optimization problem is tackled using a majorization-minimization\napproach. Finally, we develop a convex majorization technique and show that it\nenables a recursive identification method. The method learns parsimonious\npredictive models and is tested on both synthetic and real nonlinear systems.\ndoiarxiv_id1606.04366pdf_urlshttps://arxiv.org/pdf/1606.04366githubhttps://github.com/magni84/lavais_oasourcesopen-arxivtypejournaljournal_refP. Mattsson, D. Zachariah, P. Stoica, Recursive nonlinear-system\n identification using latent variables, Automatica, Volume 93, Pages 343-351,\n 2018categoriesstat.ML |
| titleVariational Calculus with Conformable Fractional Derivativesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMatheus J. LazoDelfim F. M. Torresabstract Invariant conditions for conformable fractional problems of the calculus of\nvariations under the presence of external forces in the dynamics are studied.\nDepending on the type of transformations considered, different necessary\nconditions of invariance are obtained. As particular cases, we prove fractional\nversions of Noether's symmetry theorem. Invariant conditions for fractional\noptimal control problems, using the Hamiltonian formalism, are also\ninvestigated. As an example of potential application in Physics, we show that\nwith conformable derivatives it is possible to formulate an Action Principle\nfor particles under frictional forces that is far simpler than the one obtained\nwith classical fractional derivatives.\n", "doi": "10.1109/jas.2016.7510160", "arxiv_id": "1606.07504", "pdf_urls": ["https://arxiv.org/pdf/1606.07504"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "IEEE/CAA Journal of Automatica Sinica 4 (2017), no. 2, 340--352", "categories": "math.OC math-ph math.MP"} |
| {"title": "Constrained fractional variational problems of variable order", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Dina Tavares", "Ricardo Almeida", "Delfim F. M. Torres"], "abstract": " Isoperimetric problems consist in minimizing or maximizing a cost functional\nsubject to an integral constraint. In this work, we present two fractional\nisoperimetric problems where the Lagrangian depends on a combined Caputo\nderivative of variable fractional order and we present a new variational\nproblem subject to a holonomic constraint. We establish necessary optimality\nconditions in order to determine the minimizers of the fractional problems. The\nterminal point in the cost integral, as well the terminal state, are considered\nto be free, and we obtain corresponding natural boundary conditions.\n", "doi": "10.1109/jas.2017.7510331", "arxiv_id": "1606.07512", "pdf_urls": ["https://arxiv.org/pdf/1606.07512"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "IEEE/CAA Journal of Automatica Sinica 4 (2017), no. 1, 80--88", "categories": "math.OC"} |
| {"title": "Framework for state and unknown input estimation of linear time-varying\n systems", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Peng Lu", "Erik-Jan van Kampen", "Cornelis C. de Visser", "Qiping Chu"], "abstract": " The design of unknown-input decoupled observers and filters requires the\nassumption of an existence condition in the literature. This paper addresses an\nunknown input filtering problem where the existence condition is not satisfied.\nInstead of designing a traditional unknown input decoupled filter, a\nDouble-Model Adaptive Estimation approach is extended to solve the unknown\ninput filtering problem. It is proved that the state and the unknown inputs can\nbe estimated and decoupled using the extended Double-Model Adaptive Estimation\napproach without satisfying the existence condition. Numerical examples are\npresented in which the performance of the proposed approach is compared to\nmethods from literature.\n", "doi": "10.1016/j.automatica.2016.07.009", "arxiv_id": "1606.08090", "pdf_urls": ["https://arxiv.org/pdf/1606.08090"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 73 (2016), 145-154", "categories": "cs.SY cs.IT math.IT math.ST stat.TH"} |
| {"title": "Identification of block-oriented nonlinear systems starting from linear\n approximations: A survey", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Maarten Schoukens", "Koen Tiels"], "abstract": " Block-oriented nonlinear models are popular in nonlinear system\nidentification because of their advantages of being simple to understand and\neasy to use. Many different identification approaches were developed over the\nyears to estimate the parameters of a wide range of block-oriented nonlinear\nmodels. One class of these approaches uses linear approximations to initialize\nthe identification algorithm. The best linear approximation framework and the\n$\\epsilon$-approximation framework, or equivalent frameworks, allow the user to\nextract important information about the system, guide the user in selecting\ngood candidate model structures and orders, and prove to be a good starting\npoint for nonlinear system identification algorithms. This paper gives an\noverview of the different block-oriented nonlinear models that can be\nidentified using linear approximations, and of the identification algorithms\nthat have been developed in the past. A non-exhaustive overview of the most\nimportant other block-oriented nonlinear system identification approaches is\nalso provided throughout this paper.\n", "doi": "10.1016/j.automatica.2017.06.044", "arxiv_id": "1607.01217", "pdf_urls": ["https://arxiv.org/pdf/1607.01217"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2017, vol. 85, pg. 272-292", "categories": "cs.SY"} |
| {"title": "A class of marked invariant subspaces with an application to algebraic\n Riccati equations", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Pudji Astuti", "Harald K. Wimmer"], "abstract": " Invariant subspaces of a matrix $A$ are considered which are obtained by\ntruncation of a Jordan basis of a generalized eigenspace of $A$. We\ncharacterize those subspaces which are independent of the choice of the Jordan\nbasis. An application to Hamilton matrices and algebraic Riccati equations is\ngiven.\n", "doi": null, "arxiv_id": "1607.06361", "pdf_urls": ["https://arxiv.org/pdf/1607.06361"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica J. IFAC. 42, 1503-1506 (2006)", "categories": "math.RA"} |
| {"title": "Robust Control for Signal Temporal Logic Specifications using Average\n Space Robustness", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Lars Lindemann and Dimos V. Dimarogonas"], "abstract": " Control systems that satisfy temporal logic specifications have become\nincreasingly popular due to their applicability to robotic systems. Existing\ncontrol methods, however, are computationally demanding, especially when the\nproblem size becomes too large. In this paper, a robust and computationally\nefficient model predictive control framework for signal temporal logic\nspecifications is proposed. We introduce discrete average space robustness, a\nnovel quantitative semantic for signal temporal logic, that is directly\nincorporated into the cost function of the model predictive controller. The\noptimization problem entailed in this framework can be written as a convex\nquadratic program when no disjunctions are considered and results in a robust\nsatisfaction of the specification. Furthermore, we define the predicate\nrobustness degree as a new robustness notion. Simulations of a multi-agent\nsystem subject to complex specifications demonstrate the efficacy of the\nproposed method.\n", "doi": null, "arxiv_id": "1607.07019", "pdf_urls": ["https://arxiv.org/pdf/1607.07019"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 101, pages 377-387, 2019", "categories": "cs.SY"} |
| {"title": "Identifiability of linear dynamic networks", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Harm H.M. Weerts", "Paul M.J. Van den Hof and Arne G. Dankers"], "abstract": " Dynamic networks are structured interconnections of dynamical systems\n(modules) driven by external excitation and disturbance signals. In order to\nidentify their dynamical properties and/or their topology consistently from\nmeasured data, we need to make sure that the network model set is identifiable.\nWe introduce the notion of network identifiability, as a property of a\nparameterized model set, that ensures that different network models can be\ndistinguished from each other when performing identification on the basis of\nmeasured data. Different from the classical notion of (parameter)\nidentifiability, we focus on the distinction between network models in terms of\ntheir transfer functions. For a given structured model set with a pre-chosen\ntopology, identifiability typically requires conditions on the presence and\nlocation of excitation signals, and on presence, location and correlation of\ndisturbance signals. Because in a dynamic network, disturbances cannot always\nbe considered to be of full-rank, the reduced-rank situation is also covered,\nmeaning that the number of driving white noise processes can be strictly less\nthan the number of disturbance variables. This includes the situation of having\nnoise-free nodes.\n", "doi": "10.1016/j.automatica.2017.12.013", "arxiv_id": "1609.00864", "pdf_urls": ["https://arxiv.org/pdf/1609.00864"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Vol. 89, pp. 247-258, 2018", "categories": "cs.SY"} |
| {"title": "Distributed sampled-data control of nonholonomic multi-robot systems\n with proximity networks", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Zhixin Liu", "Lin Wang", "Jinhuan Wang", "Daoyi Dong", "Xiaoming Hu"], "abstract": " This paper considers the distributed sampled-data control problem of a group\nof mobile robots connected via distance-induced proximity networks. A dwell\ntime is assumed in order to avoid chattering in the neighbor relations that may\nbe caused by abrupt changes of positions when updating information from\nneighbors. Distributed sampled-data control laws are designed based on nearest\nneighbour rules, which in conjunction with continuous-time dynamics results in\nhybrid closed-loop systems. For uniformly and independently initial states, a\nsufficient condition is provided to guarantee synchronization for the system\nwithout leaders. In order to steer all robots to move with the desired\norientation and speed, we then introduce a number of leaders into the system,\nand quantitatively establish the proportion of leaders needed to track either\nconstant or time-varying signals. All these conditions depend only on the\nneighborhood radius, the maximum initial moving speed and the dwell time,\nwithout assuming a prior properties of the neighbor graphs as are used in most\nof the existing literature.\n", "doi": "10.1016/j.automatica.2016.11.027", "arxiv_id": "1609.02174", "pdf_urls": ["https://arxiv.org/pdf/1609.02174"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 77 (2017) 170-179", "categories": "cs.SY cs.MA cs.RO"} |
| {"title": "Distributed Nash equilibrium seeking for aggregative games with coupled\n constraints", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Shu Liang", "Peng Yi", "and Yiguang Hong"], "abstract": " In this paper, we study a distributed continuous-time design for aggregative\ngames with coupled constraints in order to seek the generalized Nash\nequilibrium by a group of agents via simple local information exchange. To\nsolve the problem, we propose a distributed algorithm based on projected\ndynamics and non-smooth tracking dynamics, even for the case when the\ninteraction topology of the multi-agent network is time-varying. Moreover, we\nprove the convergence of the non-smooth algorithm for the distributed game by\ntaking advantage of its special structure and also combining the techniques of\nthe variational inequality and Lyapunov function.\n", "doi": "10.1016/j.automatica.2017.07.064", "arxiv_id": "1609.02253", "pdf_urls": ["https://arxiv.org/pdf/1609.02253"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2017, 85: 179--185", "categories": "math.OC"} |
| {"title": "Hierarchical Multistage Gaussian Signaling Games in Noncooperative\n Communication and Control Systems", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Muhammed O. Sayin", "Emrah Akyol", "and Tamer Basar"], "abstract": " We analyze in this paper finite horizon hierarchical signaling games between\n(information provider) senders and (decision maker) receivers in a dynamic\nenvironment. The underlying information evolves in time while sender and\nreceiver interact repeatedly. Different from the classical communication\n(control) models, however, the sender (sensor) and the receiver (controller)\nhave different objectives and there is a hierarchy between the players such\nthat the sender leads the game by announcing his policies beforehand. He needs\nto anticipate the reaction of the receiver and the impact of the actions on the\nhorizon while controlling the transparency of the disclosed information at each\ninteraction. With quadratic cost functions and multivariate Gaussian processes,\nevolving according to first order auto-regressive models, we show that\nmemoryless \"linear\" sender signaling rules are optimal (in the sense of\ngame-theoretic hierarchical equilibrium) within the general class of measurable\npolicies in the noncooperative communication context. In the noncooperative\ncontrol context, we also analyze the hierarchical equilibrium for linear\nsignaling rules and provide an algorithm to compute the optimal linear\nsignaling rules numerically with global optimality guarantees.\n", "doi": null, "arxiv_id": "1609.09448", "pdf_urls": ["https://arxiv.org/pdf/1609.09448"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 107, September 2019", "categories": "cs.GT math.OC"} |
| {"title": "Minimal Inputs/Outputs for Subsystems in a Networked System", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Tong Zhou"], "abstract": " Minimal input/output selection is investigated in this paper for each\nsubsystem of a networked system. Some novel sufficient conditions are derived\nrespectively for the controllability and observability of a networked system,\nas well as some necessary conditions. These conditions only depend separately\non parameters of each subsystem and its in/out-degrees. It is proven that in\norder to be able to construct a controllable/observable networked system, it is\nnecessary and sufficient that each subsystem is controllable/observable. In\naddition, both sparse and dense subsystem connections are helpful in making the\nwhole system controllable/observable. An explicit formula is given for the\nsmallest number of inputs/outputs for each subsystem required to guarantee\ncontrollability/observability of the whole system.\n", "doi": null, "arxiv_id": "1610.02192", "pdf_urls": ["https://arxiv.org/pdf/1610.02192"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Vol.94, pp.161-169, 2018", "categories": "math.OC math.DS"} |
| {"title": "Differential Inequalities in Multi-Agent Coordination and Opinion\n Dynamics Modeling", "year": 2016, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Anton V. Proskurnikov", "Ming Cao"], "abstract": " Distributed algorithms of multi-agent coordination have attracted substantial\nattention from the research community; the simplest and most thoroughly studied\nof them are consensus protocols in the form of differential or difference\nequations over general time-varying weighted graphs. These graphs are usually\ncharacterized algebraically by their associated Laplacian matrices. Network\nalgorithms with similar algebraic graph theoretic structures, called being of\nLaplacian-type in this paper, also arise in other related multi-agent control\nproblems, such as aggregation and containment control, target surrounding,\ndistributed optimization and modeling of opinion evolution in social groups. In\nspite of their similarities, each of such algorithms has often been studied\nusing separate mathematical techniques. In this paper, a novel approach is\noffered, allowing a unified and elegant way to examine many Laplacian-type\nalgorithms for multi-agent coordination. This approach is based on the analysis\nof some differential or difference inequalities that have to be satisfied by\nthe some \"outputs\" of the agents (e.g. the distances to the desired set in\naggregation problems). Although such inequalities may have many unbounded\nsolutions, under natural graphic connectivity conditions all their bounded\nsolutions converge (and even reach consensus), entailing the convergence of the\ncorresponding distributed algorithms. In the theory of differential equations\nthe absence of bounded non-convergent solutions is referred to as the\nequation's dichotomy. In this paper, we establish the dichotomy criteria of\nLaplacian-type differential and difference inequalities and show that these\ncriteria enable one to extend a number of recent results, concerned with\nLaplacian-type algorithms for multi-agent coordination and modeling opinion\nformation in social groups.\ndoi10.1016/j.automatica.2017.07.065arxiv_id1610.03373pdf_urlshttps://arxiv.org/pdf/1610.03373githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 85, November 2017, Pages 202-210categoriescs.SY cs.MA math.OC |
| titleA Data-driven Approach to Robust Control of Multivariable Systems by\n Convex Optimizationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAlireza KarimiChristoph Kammerabstract The frequency-domain data of a multivariable system in different operating\npoints is used to design a robust controller with respect to the measurement\nnoise and multimodel uncertainty. The controller is fully parametrized in terms\nof matrix polynomial functions and can be formulated as a centralized,\ndecentralized or distributed controller. All standard performance\nspecifications like $H_2$, $H_\\infty$ and loop shaping are considered in a\nunified framework for continuous- and discrete-time systems. The control\nproblem is formulated as a convex-concave optimization problem and then\nconvexified by linearization of the concave part around an initial controller.\nThe performance criterion converges monotonically to a local optimal solution\nin an iterative algorithm. The effectiveness of the method is compared with\nfixed-structure controllers using non-smooth optimization and with full-order\noptimal controllers via simulation examples. Finally, the experimental data of\na gyroscope is used to design a data-driven controller that is successfully\napplied on the real system.\ndoi10.1016/j.automatica.2017.07.063arxiv_id1610.08776pdf_urlshttps://arxiv.org/pdf/1610.08776githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica (2017)categoriesmath.OC |
| titleOn computing the distance to stability for matrices using linear\n dissipative Hamiltonian systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsNicolas Gillis and Punit Sharmaabstract In this paper, we consider the problem of computing the nearest stable matrix\nto an unstable one. We propose new algorithms to solve this problem based on a\nreformulation using linear dissipative Hamiltonian systems: we show that a\nmatrix $A$ is stable if and only if it can be written as $A = (J-R)Q$, where\n$J=-J^T$, $R \\succeq 0$ and $Q \\succ 0$ (that is, $R$ is positive semidefinite\nand $Q$ is positive definite). This reformulation results in an equivalent\noptimization problem with a simple convex feasible set. We propose three\nstrategies to solve the problem in variables $(J,R,Q)$: (i) a block coordinate\ndescent method, (ii) a projected gradient descent method, and (iii) a fast\ngradient method inspired from smooth convex optimization. These methods require\n$\\mathcal{O}(n^3)$ operations per iteration, where $n$ is the size of $A$. We\nshow the effectiveness of the fast gradient method compared to the other\napproaches and to several state-of-the-art algorithms.\ndoi10.1016/j.automatica.2017.07.047arxiv_id1611.00595pdf_urlshttps://arxiv.org/pdf/1611.00595githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 85, pp. 113-121, 2017categoriesmath.OC |
| titleRetrofit Control: Localization of Controller Design and Implementationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsTakayuki IshizakiTomonori SadamotoJun-ichi ImuraHenrik SandbergKarl Henrik Johanssonabstract In this paper, we propose a retrofit control method for stable network\nsystems. The proposed approach is a control method that, rather than an entire\nsystem model, requires a model of the subsystem of interest for controller\ndesign. To design the retrofit controller, we use a novel approach based on\nhierarchical state-space expansion that generates a higher-dimensional cascade\nrealization of a given network system. The upstream dynamics of the cascade\nrealization corresponds to an isolated model of the subsystem of interest,\nwhich is stabilized by a local controller. The downstream dynamics can be seen\nas a dynamical model representing the propagation of interference signals among\nsubsystems, the stability of which is equivalent to that of the original\nsystem. This cascade structure enables a systematic analysis of both the\nstability and control performance of the resultant closed-loop system. The\nresultant retrofit controller is formed as a cascade interconnection of the\nlocal controller and an output rectifier that rectifies an output signal of the\nsubsystem of interest so as to conform to an output signal of the isolated\nsubsystem model while acquiring complementary signals neglected in the local\ncontroller design, such as interconnection signals from neighboring subsystems.\nFinally, the efficiency of the retrofit control method is demonstrated through\nnumerical examples of power systems control and vehicle platoon control.\ndoi10.1016/j.automatica.2018.05.033arxiv_id1611.04531pdf_urlshttps://arxiv.org/pdf/1611.04531githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica (2018)categoriescs.SY math.DS |
| titleA theory of passive linear systems with no assumptionsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsTimothy H. Hughesabstract We present two linked theorems on passivity: the passive behavior theorem,\nparts 1 and 2. Part 1 provides necessary and sufficient conditions for a\ngeneral linear system, described by a set of high order differential equations,\nto be passive. Part 2 extends the positive-real lemma to include uncontrollable\nand unobservable state-space systems.\ndoi10.1016/j.automatica.2017.08.017arxiv_id1611.06140pdf_urlshttps://arxiv.org/pdf/1611.06140githubis_oasourcesopen-arxivtypejournaljournal_refHughes, T.H.: A theory of passive linear systems with no\n assumptions, Automatica, 86, 87-97 (2017)categoriescs.SY |
| titleStructural Controllability of Multi-Agent Networks: Robustness against\n Simultaneous Failuresyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsM. Amin RahimianAmir G. Aghdamabstract In this paper, structural controllability of a leader-follower multi-agent\nsystem with multiple leaders is studied from a graph-theoretic point of view.\nThe problem of preservation of structural controllability under simultaneous\nfailures in both the communication links and the agents is investigated. The\neffects of the loss of agents and communication links on the controllability of\nan information flow graph are previously studied. In this work, the\ncorresponding results are exploited to introduce some useful indices and\nimportance measures that help characterize and quantify the role of individual\nlinks and agents in the controllability of the overall network. Existing\nresults are then extended by considering the effects of losses in both links\nand agents at the same time. To this end, the concepts of joint\n(r,s)-controllability and joint t-controllability are introduced as\nquantitative measures of reliability for a multi-agent system, and their\nimportant properties are investigated. Lastly, the class of jointly critical\ndigraphs is introduced and it is stated that if a digraph is jointly critical,\nthen joint t-controllability is a necessary and sufficient condition for\nremaining controllable following the failure of any set of links and agents,\nwith cardinality less than t. Various examples are exploited throughout the\npaper to elaborate on the analytical findings.\ndoi10.1016/j.automatica.2013.06.023arxiv_id1611.10007pdf_urlshttps://arxiv.org/pdf/1611.10007githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 49, Issue 11, 2013, Pages 3149-3157categoriescs.SY cs.MA cs.RO math.CO math.DS |
| titlePower Spectrum Identification for Quantum Linear Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMatthew LevittMadalin GutaHendra I. Nurdinabstract In this paper we investigate system identification for general quantum linear\nsystems. We consider the situation where the input field is prepared as\nstationary (squeezed) quantum noise. In this regime the output field is\ncharacterised by the power spectrum, which encodes covariance of the output\nstate. We address the following two questions: (1) Which parameters can be\nidentified from the power spectrum? (2) How to construct a system realisation\nfrom the power spectrum? The power spectrum depends on the system parameters\nvia the transfer function. We show that the transfer function can be uniquely\nrecovered from the power spectrum, so that equivalent systems are related by a\nsymplectic transformation.\ndoi10.1016/j.automatica.2017.12.037arxiv_id1612.02681pdf_urlshttps://arxiv.org/pdf/1612.02681githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 90 255-262 (2018)categoriesquant-ph |
| titleWiener system identification with generalized orthonormal basis\n functionsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsKoen TielsJohan Schoukensabstract Many nonlinear systems can be described by a Wiener-Schetzen model. In this\nmodel, the linear dynamics are formulated in terms of orthonormal basis\nfunctions (OBFs). The nonlinearity is modeled by a multivariate polynomial. In\ngeneral, an infinite number of OBFs is needed for an exact representation of\nthe system. This paper considers the approximation of a Wiener system with\nfinite-order infinite impulse response dynamics and a polynomial nonlinearity.\nWe propose to use a limited number of generalized OBFs (GOBFs). The pole\nlocations, needed to construct the GOBFs, are estimated via the best linear\napproximation of the system. The coefficients of the multivariate polynomial\nare determined with a linear regression. This paper provides a convergence\nanalysis for the proposed identification scheme. It is shown that the estimated\noutput converges in probability to the exact output. Fast convergence rates, in\nthe order $O_p({N_F}^{-n_{rep}/2})$, can be achieved, with $N_F$ the number of\nexcited frequencies and $n_{rep}$ the number of repetitions of the GOBFs.\ndoi10.1016/j.automatica.2014.10.010arxiv_id1612.04558pdf_urlshttps://arxiv.org/pdf/1612.04558githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 50, Issue 12, December 2014, Pages 3147-3154categoriescs.SY |
| titleInitial estimates for Wiener-Hammerstein models using phase-coupled\n multisinesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsKoen TielsMaarten SchoukensJohan Schoukensabstract Block-oriented models are often used to model nonlinear systems. These models\nconsist of linear dynamic (L) and nonlinear static (N) sub-blocks. This paper\naddresses the generation of initial estimates for a Wiener-Hammerstein model\n(LNL cascade). While it is easy to measure the product of the two linear blocks\nusing a Gaussian excitation and linear identification methods, it is difficult\nto split the global dynamics over the individual blocks. This paper first\nproposes a well-designed multisine excitation with pairwise coupled random\nphases. Next, a modified best linear approximation is estimated on a shifted\nfrequency grid. It is shown that this procedure creates a shift of the input\ndynamics with a known frequency offset, while the output dynamics do not shift.\nThe resulting transfer function, which has complex coefficients due to the\nfrequency shift, is estimated with a modified frequency domain estimation\nmethod. The identified poles and zeros can be assigned to either the input or\noutput dynamics. Once this is done, it is shown in the literature that the\nremaining initialization problem can be solved much easier than the original\none. The method is illustrated on experimental data obtained from the\nWiener-Hammerstein benchmark system.\ndoi10.1016/j.automatica.2015.07.020arxiv_id1612.04568pdf_urlshttps://arxiv.org/pdf/1612.04568githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 60, October 2015, Pages 201-209categoriescs.SY |
| titleExponential Stability of Nonlinear Differential Repetitive Processes\n with Applications to Iterative Learning Controlyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsBerk Alt{\\i}n and Kira Bartonabstract This paper studies exponential stability properties of a class of\ntwo-dimensional (2D) systems called differential repetitive processes (DRPs).\nSince a distinguishing feature of DRPs is that the problem domain is bounded in\nthe \ direction, the notion of stability to be evaluated does not require\nthe nonlinear system defining a DRP to be stable in the typical sense. In\nparticular, we study a notion of exponential stability along the discrete\niteration dimension of the 2D dynamics, which requires the boundary data for\nthe differential pass dynamics to converge to zero as the iterations evolve.\nOur main contribution is to show, under standard regularity assumptions, that\nexponential stability of a DRP is equivalent to that of its linearized\ndynamics. In turn, exponential stability of this linearization can be readily\nverified by a spectral radius condition. The application of this result to\nPicard iterations and iterative learning control (ILC) is discussed.\nTheoretical findings are supported by a numerical simulation of an ILC\nalgorithm.\ndoi10.1016/j.automatica.2017.04.004arxiv_id1701.05665pdf_urlshttps://arxiv.org/pdf/1701.05665githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 81, July 2017, Pages 369-376categoriesmath.DS |
| titleA dynamic game approach to distributionally robust safety specifications\n for stochastic systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsInsoon Yangabstract This paper presents a new safety specification method that is robust against\nerrors in the probability distribution of disturbances. Our proposed\ndistributionally robust safe policy maximizes the probability of a system\nremaining in a desired set for all times, subject to the worst possible\ndisturbance distribution in an ambiguity set. We propose a dynamic game\nformulation of constructing such policies and identify conditions under which a\nnon-randomized Markov policy is optimal. Based on this existence result, we\ndevelop a practical design approach to safety-oriented stochastic controllers\nwith limited information about disturbance distributions. This control method\ncan be used to minimize another cost function while ensuring safety in a\nprobabilistic way. However, an associated Bellman equation involves\ninfinite-dimensional minimax optimization problems since the disturbance\ndistribution may have a continuous density. To resolve computational issues, we\npropose a duality-based reformulation method that converts the\ninfinite-dimensional minimax problem into a semi-infinite program that can be\nsolved using existing convergent algorithms. We prove that there is no duality\ngap, and that this approach thus preserves optimality. The results of numerical\ntests confirm that the proposed method is robust against distributional errors\nin disturbances, while a standard stochastic safety specification tool is not.\ndoi10.1016/j.automatica.2018.04.022arxiv_id1701.06260pdf_urlshttps://arxiv.org/pdf/1701.06260githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2018categoriesmath.OC cs.SY |
| titleDistributed sub-optimal resource allocation over weight-balanced graph\n via singular perturbationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsShu Liang and Xianlin Zeng and Yiguang Hongabstract In this paper, we consider distributed optimization design for resource\nallocation problems over weight-balanced graphs. With the help of singular\nperturbation analysis, we propose a simple sub-optimal continuous-time\noptimization algorithm. Moreover, we prove the existence and uniqueness of the\nalgorithm equilibrium, and then show the convergence with an exponential rate.\nFinally, we verify the sub-optimality of the algorithm, which can approach the\noptimal solution as an adjustable parameter tends to zero.\ndoi10.1016/j.automatica.2018.05.013arxiv_id1702.01492pdf_urlshttps://arxiv.org/pdf/1702.01492githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica,2018categoriesmath.OC |
| titleOn a class of generating vector fields for the extremum seeking problem:\n Lie bracket approximation and stability propertiesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsVictoria GrushkovskayaAlexander ZuyevChristian Ebenbauerabstract In this paper, we describe a broad class of control functions for extremum\nseeking problems. We show that it unifies and generalizes existing extremum\nseeking strategies which are based on Lie bracket approximations, and allows to\ndesign new controls with favorable properties in extremum seeking and\nvibrational stabilization tasks. The second result of this paper is a novel\napproach for studying the asymptotic behavior of extremum seeking systems. It\nprovides a constructive procedure for defining frequencies of control functions\nto ensure the practical asymptotic and exponential stability. In contrast to\nmany known results, we also prove asymptotic and exponential stability in the\nsense of Lyapunov for the proposed class of extremum seeking systems under\nappropriate assumptions on the vector fields.\ndoi10.1016/j.automatica.2018.04.024arxiv_id1703.02348pdf_urlshttps://arxiv.org/pdf/1703.02348githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica,Volume 94, Pages 151-160, 2018categoriesmath.OC |
| titleA distributed primal-dual algorithm for computation of generalized Nash\n equilibria with shared affine coupling constraints via operator splitting\n methodsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsPeng YiLacra Pavelabstract In this paper, we propose a distributed primal-dual algorithm for computation\nof a generalized Nash equilibrium (GNE) in noncooperative games over network\nsystems. In the considered game, not only each player's local objective\nfunction depends on other players' decisions, but also the feasible decision\nsets of all the players are coupled together with a globally shared affine\ninequality constraint. Adopting the variational GNE, that is the solution of a\nvariational inequality, as a refinement of GNE, we introduce a primal-dual\nalgorithm that players can use to seek it in a distributed manner. Each player\nonly needs to know its local objective function, local feasible set, and a\nlocal block of the affine constraint. Meanwhile, each player only needs to\nobserve the decisions on which its local objective function explicitly depends\nthrough the interference graph and share information related to multipliers\nwith its neighbors through a multiplier graph. Through a primal-dual analysis\nand an augmentation of variables, we reformulate the problem as finding the\nzeros of a sum of monotone operators. Our distributed primal-dual algorithm is\nbased on forward-backward operator splitting methods. We prove its convergence\nto the variational GNE for fixed step-sizes under some mild assumptions. Then a\ndistributed algorithm with inertia is also introduced and analyzed for\nvariational GNE seeking. Finally, numerical simulations for network Cournot\ncompetition are given to illustrate the algorithm efficiency and performance.\ndoi10.1016/j.automatica.2019.01.008arxiv_id1703.05388pdf_urlshttps://arxiv.org/pdf/1703.05388githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol 102, pp. 111 -121, 2019categoriesmath.OC cs.GT cs.SY |
| titleD-Optimal Input Design for Nonlinear FIR-type Systems:A Dispersion-based\n Approachyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAlexander De CockMichel GeversJohan Schoukensabstract Optimal input design is an important step of the identification process in\norder to reduce the model variance. In this work a D-optimal input design\nmethod for finite-impulse-response-type nonlinear systems is presented. The\noptimization of the determinant of the Fisher information matrix is expressed\nas a convex optimization problem. This problem is then solved using a\ndispersion-based optimization scheme, which is easy to implement and converges\nmonotonically to the optimal solution. Without constraints, the optimal design\ncannot be realized as a time sequence. By imposing that the design should lie\nin the subspace described by a symmetric and non-overlapping set, a realizable\ndesign is found. A graph-based method is used in order to find a time sequence\nthat realizes this optimal constrained design. These methods are illustrated on\na numerical example of which the results are thoroughly discussed. Additionally\nthe computational speed of the algorithm is compared with the general convex\noptimizer cvx.\ndoiarxiv_id1703.08401pdf_urlshttps://arxiv.org/pdf/1703.08401githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Vol. 73, pages 88-100, November 2016categoriescs.SY |
| titleUseful redundancy in parameter and time delay estimation for\n continuous-time modelsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsHuong HaJames S. WelshMazen Alamirabstract This paper proposes an algorithm to estimate the parameters, including time\ndelay, of continuous time systems based on instrumental variable identification\nmethods. To overcome the multiple local minima of the cost function associated\nwith the estimation of a time delay system, we utilise the useful redundancy\ntechnique. Specifically, the cost function is filtered through a set of\nlow-pass filters to improve convexity with the useful redundancy technique\nexploited to achieve convergence to the global minimum of the optimization\nproblem. Numerical examples are presented to demonstrate the effectiveness of\nthe proposed algorithm.\ndoi10.1016/j.automatica.2018.06.023arxiv_id1703.09344pdf_urlshttps://arxiv.org/pdf/1703.09344githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 95, 455-462 (2018)categoriesmath.ST stat.TH |
| titleAn analysis of the SPARSEVA estimate for the finite sample data caseyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsHuong HaJames S. WelshCristian R. RojasBo Wahlbergabstract In this paper, we develop an upper bound for the SPARSEVA (SPARSe Estimation\nbased on a VAlidation criterion) estimation error in a general scheme, i.e.,\nwhen the cost function is strongly convex and the regularized norm is\ndecomposable for a pair of subspaces. We show how this general bound can be\napplied to a sparse regression problem to obtain an upper bound for the\ntraditional SPARSEVA problem. Numerical results are used to illustrate the\neffectiveness of the suggested bound.\ndoi10.1016/j.automatica.2018.06.046arxiv_id1703.09351pdf_urlshttps://arxiv.org/pdf/1703.09351githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 96, 141-149, 2018categoriesmath.ST stat.TH |
| titleA Distributed Nash Equilibrium Seeking in Networked Graphical Gamesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsFarzad Salehisadaghiani and Lacra Pavelabstract This paper considers a distributed gossip approach for finding a Nash\nequilibrium in networked games on graphs. In such games a player's cost\nfunction may be affected by the actions of any subset of players. An\ninterference graph is employed to illustrate the partially-coupled cost\nfunctions and the asymmetric information requirements. For a given interference\ngraph, network communication between players is considered to be limited. A\ngeneralized communication graph is designed so that players exchange only their\nrequired information. An algorithm is designed whereby players, with possibly\npartially-coupled cost functions, make decisions based on the estimates of\nother players' actions obtained from local neighbors. It is shown that this\nchoice of communication graph guarantees that all players' information is\nexchanged after sufficiently many iterations. Using a set of standard\nassumptions on the cost functions, the interference and the communication\ngraphs, almost sure convergence to a Nash equilibrium is proved for diminishing\nstep sizes. Moreover, the case when the cost functions are not known by the\nplayers is investigated and a convergence proof is presented for diminishing\nstep sizes. The effect of the second largest eigenvalue of the expected\ncommunication matrix on the convergence rate is quantified. The trade-off\nbetween parameters associated with the communication graph and the ones\nassociated with the interference graph is illustrated. Numerical results are\npresented for a large-scale networked game.\n", "doi": "10.1016/j.automatica.2017.09.016", "arxiv_id": "1703.09765", "pdf_urls": ["https://arxiv.org/pdf/1703.09765"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 87, pp. 17 - 24, 2018", "categories": "cs.GT cs.SY math.OC"} |
| {"title": "Boundary feedback stabilization of a flexible wing model under unsteady\n aerodynamic loads", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Hugo Lhachemi", "David Saussi\\'eand Guchuan Zhuabstract This paper addresses the boundary stabilization of a flexible wing model,\nboth in bending and twisting displacements, under unsteady aerodynamic loads,\nand in presence of a store. The wing dynamics is captured by a distributed\nparameter system as a coupled Euler-Bernoulli and Timoshenko beam model. The\nproblem is tackled in the framework of semigroup theory, and a Lyapunov-based\nstability analysis is carried out to assess that the system energy, as well as\nthe bending and twisting displacements, decay exponentially to zero. The\neffectiveness of the proposed boundary control scheme is evaluated based on\nsimulations.\ndoi10.1016/j.automatica.2018.07.029arxiv_id1703.10182pdf_urlshttps://arxiv.org/pdf/1703.10182githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 97, November 2018, Pages 73-81categoriesmath.OC cs.SY |
| titleRisk-averse model predictive controlyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsPantelis SopasakisDomagoj HercegAlberto BemporadPanagiotis\n Patrinosabstract Risk-averse model predictive control (MPC) offers a control framework that\nallows one to account for ambiguity in the knowledge of the underlying\nprobability distribution and unifies stochastic and worst-case MPC. In this\npaper we study risk-averse MPC problems for constrained nonlinear Markovian\nswitching systems using generic cost functions, and derive Lyapunov-type\nrisk-averse stability conditions by leveraging the properties of risk-averse\ndynamic programming operators. We propose a controller design procedure to\ndesign risk-averse stabilizing terminal conditions for constrained nonlinear\nMarkovian switching systems. Lastly, we cast the resulting risk-averse optimal\ncontrol problem in a favorable form which can be solved efficiently and thus\ndeems risk-averse MPC suitable for applications.\ndoi10.1016/j.automatica.2018.11.022arxiv_id1704.00342pdf_urlshttps://arxiv.org/pdf/1704.00342githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 100, pp. 281-288, Feb 2019categoriesmath.OC |
| titleMinimal Controllability of Conjunctive Boolean Networks is NP-Completeyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsEyal Weiss and Michael Margaliot and Guy Evenabstract Given a conjunctive Boolean network (CBN) with $n$ state-variables, we\nconsider the problem of finding a minimal set of state-variables to directly\naffect with an input so that the resulting conjunctive Boolean control network\n(CBCN) is controllable. We give a necessary and sufficient condition for\ncontrollability of a CBCN; an $O(n^2)$-time algorithm for testing\ncontrollability; and prove that nonetheless the minimal controllability problem\nfor CBNs is NP-hard.\ndoi10.1016/j.automatica.2018.02.014arxiv_id1704.07291pdf_urlshttps://arxiv.org/pdf/1704.07291githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 92 (2018): 56-62categoriescs.DS cs.CC cs.SY math.OC q-bio.MN |
| titlePolicy Iterations for Reinforcement Learning Problems in Continuous Time\n and Space -- Fundamental Theory and Methodsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsJaeyoung Lee and Richard S. Suttonabstract Policy iteration (PI) is a recursive process of policy evaluation and\nimprovement for solving an optimal decision-making/control problem, or in other\nwords, a reinforcement learning (RL) problem. PI has also served as the\nfundamental for developing RL methods. In this paper, we propose two PI\nmethods, called differential PI (DPI) and integral PI (IPI), and their\nvariants, for a general RL framework in continuous time and space (CTS), where\nthe environment is modeled by a system of ordinary differential equations\n(ODEs). The proposed methods inherit the current ideas of PI in classical RL\nand optimal control and theoretically support the existing RL algorithms in\nCTS: TD-learning and value-gradient-based (VGB) greedy policy update. We also\nprovide case studies including 1) discounted RL and 2) optimal control tasks.\nFundamental mathematical properties -- admissibility, uniqueness of the\nsolution to the Bellman equation (BE), monotone improvement, convergence, and\noptimality of the solution to the Hamilton-Jacobi-Bellman equation (HJBE) --\nare all investigated in-depth and improved from the existing theory, along with\nthe general and case studies. Finally, the proposed ones are simulated with an\ninverted-pendulum model and their model-based and partially model-free\nimplementations to support the theory and further investigate them beyond.\ndoi10.1016/j.automatica.2020.109421arxiv_id1705.03520pdf_urlshttps://arxiv.org/pdf/1705.03520githubhttps://github.com/JaeyoungLee-UoA/PIs-for-RL-Problems-in-CTSis_oasourcesopen-arxivtypejournaljournal_refAutomatica vol. 126, 109421 (2021)categoriescs.AI cs.LG cs.SY eess.SY |
| titleAn Efficient LQR Design for Discrete-Time Linear Periodic System Based\n on a Novel Lifting Methodyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsYaguang Yangabstract This paper proposes a novel lifting method which converts the standard\ndiscrete-time linear periodic system to an augmented linear time-invariant\nsystem. The linear quadratic optimal control is then based on the solution of\nthe discrete-time algebraic Riccati equation associated with the augmented\nlinear time-invariant model. An efficient algorithm for solving the Riccati\nequation is derived by using the special structure of the augmented linear\ntime-invariant system. It is shown that the proposed method is very efficient\ncompared to the ones that use algorithms for discrete-time periodic algebraic\nRiccati equation. The efficiency and effectiveness of the proposed algorithm is\ndemonstrated by the simulation test for the design problem of spacecraft\nattitude control using magnetic torques.\ndoi10.1016/j.automatica.2017.10.019arxiv_id1705.04617pdf_urlshttps://arxiv.org/pdf/1705.04617githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2018categoriesmath.OC |
| titleAn Exponential Quantum Projection Filter for Open Quantum Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsQing GaoGuofeng Zhang and Ian R. Petersenabstract An approximate exponential quantum projection filtering scheme is developed\nfor a class of open quantum systems described by Hudson- Parthasarathy quantum\nstochastic differential equations, aiming to reduce the computational burden\nassociated with online calculation of the quantum filter. By using a\ndifferential geometric approach, the quantum trajectory is constrained in a\nfinite-dimensional differentiable manifold consisting of an unnormalized\nexponential family of quantum density operators, and an exponential quantum\nprojection filter is then formulated as a number of stochastic differential\nequations satisfied by the finite-dimensional coordinate system of this\nmanifold. A convenient design of the differentiable manifold is also presented\nthrough reduction of the local approximation errors, which yields a\nsimplification of the quantum projection filter equations. It is shown that the\ncomputational cost can be significantly reduced by using the quantum projection\nfilter instead of the quantum filter. It is also shown that when the quantum\nprojection filtering approach is applied to a class of open quantum systems\nthat asymptotically converge to a pure state, the input-to-state stability of\nthe corresponding exponential quantum projection filter can be established.\nSimulation results from an atomic ensemble system example are provided to\nillustrate the performance of the projection filtering scheme. It is expected\nthat the proposed approach can be used in developing more efficient quantum\ncontrol methods.\ndoiarxiv_id1705.09114pdf_urlshttps://arxiv.org/pdf/1705.09114githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica-2018categoriesmath-ph math.MP quant-ph |
| titleSparse and Constrained Stochastic Predictive Control for Networked\n Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsPrabhat K. MishraDebasish ChatterjeeDaniel E. Quevedoabstract This article presents a novel class of control policies for networked control\nof Lyapunov-stable linear systems with bounded inputs. The control channel is\nassumed to have i.i.d. Bernoulli packet dropouts and the system is assumed to\nbe affected by additive stochastic noise. Our proposed class of policies is\naffine in the past dropouts and saturated values of the past disturbances. We\nfurther consider a regularization term in a quadratic performance index to\npromote sparsity in control. We demonstrate how to augment the underlying\noptimization problem with a constant negative drift constraint to ensure\nmean-square boundedness of the closed-loop states, yielding a convex quadratic\nprogram to be solved periodically online. The states of the closed-loop plant\nunder the receding horizon implementation of the proposed class of policies are\nmean square bounded for any positive bound on the control and any non-zero\nprobability of successful transmission.\ndoi10.1016/j.automatica.2017.09.013arxiv_id1706.01792pdf_urlshttps://arxiv.org/pdf/1706.01792githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Vol. 87, pp. 40-51, 2018categoriesmath.OC |
| titleRecovering Markov Models from Closed-Loop Datayearvenue_abbrAutomaticavenue_nameAutomaticaauthorsJonathan P. EpperleinRobert ShortenSergiy Zhukabstract Situations in which recommender systems are used to augument decision making\nare becoming prevalent in many application domains. Almost always, these\nprediction tools (recommenders) are created with a view to affecting\nbehavioural change. Clearly, successful applications actuating behavioural\nchange, affect the original model underpinning the predictor, leading to an\ninconsistency. This feedback loop is often not considered in standard so-called\nBig Data learning techniques which rely upon machine learning/statistical\nlearning machinery. The objective of this paper is to develop tools that\nrecover unbiased user models in the presence of recommenders. More\nspecifically, we assume that we observe a time series which is a trajectory of\na Markov chain ${R}$ modulated by another Markov chain ${S}$, i.e. the\ntransition matrix of ${R}$ is unknown and depends on the current state of\n${S}$. The transition matrix of the latter is also unknown. In other words, at\neach time instant, ${S}$ selects a transition matrix for ${R}$ within a given\nset which consists of known and unknown matrices. The state of ${S}$, in turn,\ndepends on the current state of ${R}$ thus introducing a feedback loop. We\npropose an Expectation-Maximization (EM) type algorithm, which estimates the\ntransition matrices of ${S}$ and ${R}$. Experimental results are given to\ndemonstrate the efficacy of the approach.\ndoi10.1016/j.automatica.2019.01.022arxiv_id1706.06359pdf_urlshttps://arxiv.org/pdf/1706.06359githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 103, 2019, Pages 116-125, ISSN 0005-1098categoriesmath.OC |
| titleDistributed Nash Equilibrium Seeking under Partial-Decision Information\n via the Alternating Direction Method of Multipliersyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsFarzad SalehisadaghianiWei Shi and Lacra Pavelabstract In this paper, we consider the problem of finding a Nash equilibrium in a\nmulti-player game over generally connected networks. This model differs from a\nconventional setting in that players have partial information on the actions of\ntheir opponents and the communication graph is not necessarily the same as the\nplayers' cost dependency graph. We develop a relatively fast algorithm within\nthe framework of inexact-ADMM, based on local information exchange between the\nplayers. We prove its convergence to Nash equilibrium for fixed step-sizes and\nanalyze its convergence rate. Numerical simulations illustrate its benefits\nwhen compared to a consensus-based gradient type algorithm with diminishing\nstep-sizes.\n", "doi": "10.1016/j.automatica.2019.01.012", "arxiv_id": "1707.01965", "pdf_urls": ["https://arxiv.org/pdf/1707.01965"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol 103, pp. 27 - 35, 2019", "categories": "math.OC"} |
| {"title": "Contour and Centreline Tracking of Vessels from Angiograms using the\n Classical Image Processing Techniques", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Tache Irina Andra"], "abstract": " This article deals with the problem of vessel edge and centerline detection\nusing classical image processing techniques due to their simpleness and\neasiness to be implemented. The method is divided into four steps: the vessel\nenhancement which implies a non-linear filtering proposed by Frangi, the\nthresholding using Otsu method and the contour detection using the Canny edge\ndetector due to its good performances for the small vessels and the\nmorphological skeletonisation. The algorithms are tested on real data collected\nfrom a cardiac catheterism laboratory and it is accurate for images with good\nspatial resolution (512*512). The output image can be used for further\nprocessing in order to find the vessel length or its radius.\n", "doi": null, "arxiv_id": "1707.03710", "pdf_urls": ["https://arxiv.org/pdf/1707.03710"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Buletinul Institutului Politehnic din Iasi, Sectia Automatica si\n Calculatoare, Vol. 62(66), No. 3, pp. 49-60, July 2016, ISSN 1220-2169", "categories": "cs.CV"} |
| {"title": "Quantum gate identification: error analysis, numerical results and\n optical experiment", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Yuanlong Wang", "Qi Yin", "Daoyi Dong", "Bo Qi", "Ian R. Petersen", "Zhibo Hou", "Hidehiro Yonezawa", "Guo-Yong Xiang"], "abstract": " The identification of an unknown quantum gate is a significant issue in\nquantum technology. In this paper, we propose a quantum gate identification\nmethod within the framework of quantum process tomography. In this method, a\nseries of pure states are inputted to the gate and then a fast state tomography\non the output states is performed and the data are used to reconstruct the\nquantum gate. Our algorithm has computational complexity $O(d^3)$ with the\nsystem dimension $d$. The algorithm is compared with maximum likelihood\nestimation method for the running time, which shows the efficiency advantage of\nour method. An error upper bound is established for the identification\nalgorithm and the robustness of the algorithm against the purity of input\nstates is also tested. We perform quantum optical experiment on single-qubit\nHadamard gate to verify the effectiveness of the identification algorithm.\n", "doi": "10.1016/j.automatica.2018.12.011", "arxiv_id": "1707.06039", "pdf_urls": ["https://arxiv.org/pdf/1707.06039"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2019, Vol. 101, pp. 269-279", "categories": "quant-ph cs.SY"} |
| {"title": "Stability analysis of a system coupled to a heat equation", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Lucie Baudouin (LAAS-MAC)", "Alexandre Seuret (LAAS-MAC)", "Fr\\'ed\\'eric\n Gouaisbaut (LAAS-MAC)"], "abstract": " As a first approach to the study of systems coupling finite and infinite\ndimensional natures, this article addresses the stability of a system of\nordinary differential equations coupled with a classic heat equation using a\nLyapunov functional technique. Inspired from recent developments in the area of\ntime delay systems, a new methodology to study the stability of such a class of\ndistributed parameter systems is presented here. The idea is to use a\npolynomial approximation of the infinite dimensional state of the heat equation\nin order to build an enriched energy functional. A well known efficient\nintegral inequality (Bessel inequality) will allow to obtain stability\nconditions expressed in terms of linear matrix inequalities. We will eventually\ntest our approach on academic examples in order to illustrate the efficiency of\nour theoretical results.\n", "doi": null, "arxiv_id": "1707.07967", "pdf_urls": ["https://arxiv.org/pdf/1707.07967"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Elsevier, 2019, 99, pp.195-202", "categories": "math.AP"} |
| {"title": "Global Finite-Time Attitude Consensus of Leader-Following Spacecraft\n Systems Based on Distributed Observers", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Haichao Gui and Anton H. J. de Ruiter"], "abstract": " This paper addresses the leader-following attitude consensus problem for a\ngroup of spacecraft when at least one follower can access the leader's attitude\nand velocity relative to the inertial space. A nonlinear distributed observer\nis designed to estimate the leader's states for each follower. The observer\npossesses one important and novel feature of keeping attitude and angular\nvelocity estimation errors on second-order sliding modes, and thus provides\nfinite-time convergent estimates for each follower. Further, quaternion-based\nhybrid homogeneous controllers recently developed for single spacecraft are\nextended and then applied, by establishing a separation principle with the\nproposed observer, to track the leader's attitude motion. As a result, global\nfinite-time attitude consensus is achieved on the entire attitude manifold,\nwith either full-state measurements or attitude-only measurements, as long as\nthe network topology among the followers is undirected and connected. Numerical\nsimulations are presented to demonstrate the performance of the proposed\nmethods.\ndoi10.1016/j.automatica.2018.01.037arxiv_id1707.08076pdf_urlshttps://arxiv.org/pdf/1707.08076githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2018, 91: 225~232categoriesmath.OC |
| titleCooperative Global Robust Output Regulation for a Class of Nonlinear\n Multi-Agent Systems by Distributed Event-Triggered Controlyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsWei Liu and Jie Huangabstract This paper studies the event-triggered cooperative global robust output\nregulation problem for a class of nonlinear multi-agent systems via a\ndistributed internal model design. We show that our problem can be solved\npractically in the sense that the ultimate bound of the tracking error can be\nmade arbitrarily small by adjusting a design parameter in the proposed\nevent-triggered mechanism. Our result offers a few new features. First, our\ncontrol law is robust against both external disturbances and parameter\nuncertainties, which are allowed to belong to some arbitrarily large prescribed\ncompact sets. Second, the nonlinear functions in our system do not need to\nsatisfy the global Lipchitz condition. Thus our systems are general enough to\ninclude some benchmark nonlinear systems that cannot be handled by existing\napproaches. Finally, our control law is a specific distributed output-based\nevent-triggered control law, which lends itself to a direct digital\nimplementation.\ndoi10.1016/j.automatica.2018.03.062arxiv_id1707.08421pdf_urlshttps://arxiv.org/pdf/1707.08421githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 93, pp. 138-148, July 2018categoriesmath.OC |
| titleAn Optimal Control Formulation of Pulse-Based Control Using Koopman\n Operatoryearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAivar SootlaAlexandre Mauroy and Damien Ernstabstract In many applications, and in systems/synthetic biology, in particular, it is\ndesirable to compute control policies that force the trajectory of a bistable\nsystem from one equilibrium (the initial point) to another equilibrium (the\ntarget point), or in other words to solve the switching problem. It was\nrecently shown that, for monotone bistable systems, this problem admits\neasy-to-implement open-loop solutions in terms of temporal pulses (i.e., step\nfunctions of fixed length and fixed magnitude). In this paper, we develop this\nidea further and formulate a problem of convergence to an equilibrium from an\narbitrary initial point. We show that this problem can be solved using a static\noptimization problem in the case of monotone systems. Changing the initial\npoint to an arbitrary state allows to build closed-loop, event-based or\nopen-loop policies for the switching/convergence problems. In our derivations\nwe exploit the Koopman operator, which offers a linear infinite-dimensional\nrepresentation of an autonomous nonlinear system. One of the main advantages of\nusing the Koopman operator is the powerful computational tools developed for\nthis framework. Besides the presence of numerical solutions, the\nswitching/convergence problem can also serve as a building block for solving\nmore complicated control problems and can potentially be applied to\nnon-monotone systems. We illustrate this argument on the problem of\nsynchronizing cardiac cells by defibrillation. Potentially, our approach can be\nextended to problems with different parametrizations of control signals since\nthe only fundamental limitation is the finite time application of the control\nsignal.\ndoi10.1016/j.automatica.2018.01.036arxiv_id1707.08462pdf_urlshttps://arxiv.org/pdf/1707.08462githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica Volume 91, May 2018, Pages 217-224categoriesmath.OC cs.SY |
| titleParametric identification of parallel Wiener-Hammerstein systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMaarten SchoukensAnna MarconatoRik PintelonGerd VandersteenYves Rolainabstract Block-oriented nonlinear models are popular in nonlinear modeling because of\ntheir advantages to be quite simple to understand and easy to use. To increase\nthe flexibility of single branch block-oriented models, such as Hammerstein,\nWiener, and Wiener-Hammerstein models, parallel block-oriented models can be\nconsidered. This paper presents a method to identify parallel\nWiener-Hammerstein systems starting from input-output data only. In the first\nstep, the best linear approximation is estimated for different input excitation\nlevels. In the second step, the dynamics are decomposed over a number of\nparallel orthogonal branches. Next, the dynamics of each branch are partitioned\ninto a linear time invariant subsystem at the input and a linear time invariant\nsubsystem at the output. This is repeated for each branch of the model. The\nstatic nonlinear part of the model is also estimated during this step. The\nconsistency of the proposed initialization procedure is proven. The method is\nvalidated on real-world measurements using a custom built parallel\nWiener-Hammerstein test system.\ndoi10.1016/j.automatica.2014.10.105arxiv_id1708.06543pdf_urlshttps://arxiv.org/pdf/1708.06543githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2015, vol. 51, pg. 111 - 122categoriescs.SY |
| titleEvent-Triggered Stabilization of Nonlinear Systems with Time-Varying\n Sensing and Actuation Delayyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsErfan NozariPavankumar TallapragadaJorge Cort\\'es"], "abstract": " This paper studies the problem of stabilization of a nonlinear system with\ntime-varying delays in both sensing and actuation using event-triggered\ncontrol. Our proposed strategy seeks to opportunistically minimize the number\nof control updates while guaranteeing stabilization and builds on predictor\nfeedback to compensate for arbitrarily large known time-varying delays. We\nestablish, using a Lyapunov approach, the global asymptotic stability of the\nclosed-loop system as long as the open-loop system is globally input-to-state\nstabilizable in the absence of time delays and sampling. We further prove that\nthe proposed event-triggered law has inter-event times that are uniformly lower\nbounded and hence does not exhibit Zeno behavior. For the particular case of a\nstabilizable linear system, we show global exponential stability of the\nclosed-loop system and analyze the trade-off between the rate of exponential\nconvergence and a bound on the sampling frequency. We illustrate these results\nin simulation and also examine the properties of the proposed event-triggered\nstrategy beyond the class of systems for which stabilization can be guaranteed.\n", "doi": null, "arxiv_id": "1709.02055", "pdf_urls": ["https://arxiv.org/pdf/1709.02055"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Vol. 113, March 2020, p. 108754", "categories": "cs.SY math.DS"} |
| {"title": "Switching and Information Exchange in Compressed Estimation of Coupled\n High Dimensional Processes", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Karan Narula and Jose Guivant"], "abstract": " Compressed Estimation approaches, such as the Generalised Compressed Kalman\nFilter (GCKF), reduce the computational cost and complexity of high dimensional\nand high frequency data assimilation problems; usually without sacrificing\noptimality. Configured using adequate cores, such as the Unscented Kalman\nFilter (UKF), the GCKF could also treat certain non-linear cases. However, the\napplication of a compressed estimation process is limited to a class of\nproblems which inherently allow the estimation process to be divided, at\ncertain intervals of time, in a subset of lower dimensional problems. This\nlimitation prohibits applying the compressing techniques for estimating densely\ncoupled high dimensional processes. However, those limitations can be overcome\nby applying proper techniques. In this paper, the concept of subsystem\nswitching, and information exchange architecture, namely Exploiting Local\nStatistical Dependency (ELSD), has been derived and explored, for allowing\ncompressed estimators to mimic optimal full Gaussian estimators. The\nperformances of the methods have been verified through its application in\nsolving usual types of linear Stochastic Partial Differential Equations\n(SPDEs). The computational advantages of using the proposed techniques have\nalso been highlighted with recommendation of its usage over the full filter\nwhen dealing with high dimensional and high frequency data assimilation.\n", "doi": "10.1016/j.automatica.2018.10.018", "arxiv_id": "1709.09041", "pdf_urls": ["https://arxiv.org/pdf/1709.09041"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 99 (2019) 149-156", "categories": "cs.SY math.OC math.PR"} |
| {"title": "Prediction error identification of linear dynamic networks with\n rank-reduced noise", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Harm H.M. Weerts", "Paul M.J. Van den Hof and Arne G. Dankers"], "abstract": " Dynamic networks are interconnected dynamic systems with measured node\nsignals and dynamic modules reflecting the links between the nodes. We address\nthe problem of \\red{identifying a dynamic network with known topology, on the\nbasis of measured signals}, for the situation of additive process noise on the\nnode signals that is spatially correlated and that is allowed to have a\nspectral density that is singular. A prediction error approach is followed in\nwhich all node signals in the network are jointly predicted. The resulting\njoint-direct identification method, generalizes the classical direct method for\nclosed-loop identification to handle situations of mutually correlated noise on\ninputs and outputs. When applied to general dynamic networks with rank-reduced\nnoise, it appears that the natural identification criterion becomes a weighted\nLS criterion that is subject to a constraint. This constrained criterion is\nshown to lead to maximum likelihood estimates of the dynamic network and\ntherefore to minimum variance properties, reaching the Cramer-Rao lower bound\nin the case of Gaussian noise.\n", "doi": "10.1016/j.automatica.2018.09.033", "arxiv_id": "1711.06369", "pdf_urls": ["https://arxiv.org/pdf/1711.06369"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Vol. 98, pp. 256-268, December 2018", "categories": "cs.SY"} |
| {"title": "A New Approach To Estimate The Collision Probability For Automotive\n Applications", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Richard Altendorfer and Christoph Wilkmann"], "abstract": " We revisit the computation of a probability of collision in the context of\nautomotive collision avoidance (also referred to as conflict detection in other\ncontexts). After reviewing existing approaches to the definition and\ncomputation of a collision probability we argue that the question \"What is the\nprobability of collision within the next three seconds?\" can be answered on the\nbasis of a collision probability rate.\n Using results on level crossings for vector stochastic processes we derive a\ngeneral expression for the upper bound of the distribution of the collision\nprobability rate. This expression is valid for arbitrary prediction models\nincluding process noise.\n We demonstrate in several examples that distributions obtained by large-scale\nMonte-Carlo simulations obey this bound and in many cases approximately\nsaturate the bound. We derive an approximation for the distribution of the\ncollision probability rate that can be computed on an embedded platform. An\nupper bound of the probability of collision is then obtained by one-dimensional\nnumerical integration over the time period of interest.\n A straightforward application of this method applies to the collision of an\nextended object with a second point-like object. Using an abstraction of the\nsecond object by salient points of its boundary we propose an application of\nthis method to two extended objects with arbitrary orientation.\n Finally, the distribution of the collision probability rate is compared to\napproximations of time-to-collision distributions for one-dimensional motions\nthat have been obtained previously.\n", "doi": "10.1016/j.automatica.2021.109497", "arxiv_id": "1711.07060", "pdf_urls": ["https://arxiv.org/pdf/1711.07060"], "github": ["https://github.com/ken-power/SensorFusionND-3D-Object-Tracking"], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 127, May 2021, 109497", "categories": "eess.SY cs.SY"} |
| {"title": "Finite Horizon Robustness Analysis of LTV Systems Using Integral Quadratic Constraints", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Peter Seiler and Robert Moore and Chris Meissen and Murat Arcak and Andrew Packard"], "abstract": "The goal of this paper is to assess the robustness of an uncertain linear time-varying (LTV) system on a finite time horizon. The uncertain system is modeled as a connection of a known LTV system and a perturbation. The input/output behavior of the perturbation is described by time-domain, integral quadratic constraints (IQCs). Typical notions of robustness, e.g. nominal stability and gain/phase margins, can be insufficient for finite-horizon analysis. Instead, this paper focuses on robust induced gains and bounds on the reachable set of states. Sufficient conditions to compute robust performance bounds are formulated using dissipation inequalities and IQCs. The analysis conditions are provided in two equivalent forms as Riccati differential equations and differential linear matrix inequalities. A computational approach is provided that leverages both forms of the analysis conditions. The approach is demonstrated with two examples", "doi": "10.1016/j.automatica.2018.11.009", "arxiv_id": "1711.07248", "pdf_urls": ["https://arxiv.org/pdf/1711.07248"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, volume 100, pp 135-143, 2019", "categories": "cs.SY math.OC"} |
| {"title": "Observers for a non-Lipschitz triangular form", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Pauline Bernard (CAS)", "Laurent Praly (CAS)", "Vincent Andrieu (LAGEP)"], "abstract": " We address the problem of designing an observer for triangular non locally\nLipschitz dynamical systems. We show the convergence with an arbitrary small\nerror of the classical high gain observer in presence of nonlinearities\nverifying some H{\\\"o}lder-like condition. Also, for the case when this\nH{\\\"o}lder condition is not verified, we propose a novel cascaded high gain\nobserver. Under slightly more restrictive assumptions, we prove the convergence\nof a homogeneous observer and of its cascaded version with the help of an\nexplicit Lyapunov function.\n", "doi": "10.1016/j.automatica.2017.04.054", "arxiv_id": "1711.09727", "pdf_urls": ["https://arxiv.org/pdf/1711.09727"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Elsevier, 2017, 82, pp.301-313", "categories": "math.OC"} |
| {"title": "On the triangular canonical form for uniformly observable controlled\n systems", "year": 2017, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Pauline Bernard (CAS)", "Laurent Praly (CAS)", "Vincent Andrieu (LAGEP)", "Hassan Hammouri (LAGEP)"], "abstract": " We study controlled systems which are uniformly observable and differentially\nobservable with an order larger than the system state dimension. We establish\nthat they may be transformed into a (partial) triangular canonical form but\nwith possibly non locally Lipschitz functions. We characterize the points where\nthis Lipschitzness may be lost and investigate the link with uniform\ninfinitesimal observability.\n", "doi": "10.1016/j.automatica.2017.07.034", "arxiv_id": "1711.10237", "pdf_urls": ["https://arxiv.org/pdf/1711.10237"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Elsevier, 2017, 85, pp.293 - 300", "categories": "math.OC math.DG"} |
| {"title": "Initialization-free Privacy-guaranteed Distributed Algorithm for\n Economic Dispatch Problem", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Hyeonjun Yun", "Hyungbo Shim", "Hyo-Sung Ahn"], "abstract": " This paper considers the economic dispatch problem for a network of power\ngenerators and customers. In particular, our aim is to minimize the total\ngeneration cost under the power supply-demand balance and the individual\ngeneration capacity constraints. This problem is solved in a distributed\nmanner, i.e., a dual gradient-based continuous-time distributed algorithm is\nproposed in which only a single dual variable is communicated with the\nneighbors and no private information of the node is disclosed. The proposed\nalgorithm is simple and no specific initialization is necessary, and this in\nturn allows on-line change of network structure, demand, generation\nconstraints, and even the participating nodes. The algorithm also exhibits a\nspecial behavior when the problem becomes infeasible so that each node can\ndetect over-demand or under-demand situation of the power network. Simulation\nresults on IEEE 118 bus system confirm robustness against variations in power\ngrids.\n", "doi": "10.1016/j.automatica.2018.12.033", "arxiv_id": "1801.01658", "pdf_urls": ["https://arxiv.org/pdf/1801.01658"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 102, pp. 86-93, April 2019", "categories": "cs.SY"} |
| {"title": "Exploiting structure of chance constrained programs via submodularity", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Damian Frick and Pier Giuseppe Sessa and Tony A. Wood and Maryam\n Kamgarpour"], "abstract": " We introduce a novel approach to reduce the computational effort of solving\nmixed-integer convex chance constrained programs through the scenario approach.\nInstead of reducing the number of required scenarios, we directly minimize the\ncomputational cost of the scenario program. We exploit the problem structure by\nefficiently partitioning the constraint function and considering a multiple\nchance constrained program that gives the same probabilistic guarantees as the\noriginal single chance constrained problem. We formulate the problem of finding\nthe optimal partition, a partition achieving the lowest computational cost, as\nan optimization problem with nonlinear objective and combinatorial constraints.\nBy using submodularity of the support rank of a set of constraints, we propose\na polynomial-time algorithm to find suboptimal solutions to this partitioning\nproblem and we give approximation guarantees for special classes of cost\nmetrics. We illustrate that the resulting computational cost savings can be\narbitrarily large and demonstrate our approach on two case studies from\nproduction and multi-agent planning.\n", "doi": "10.1016/j.automatica.2019.03.027", "arxiv_id": "1801.03258", "pdf_urls": ["https://arxiv.org/pdf/1801.03258"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica Volume 105, July 2019, Pages 89-95", "categories": "math.OC"} |
| {"title": "Control for Networked Control Systems with Remote and Local Controllers\n over Unreliable Communication Channel", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Xiao Liang", "Juanjuan Xu"], "abstract": " This paper is concerned with the problems of optimal control and\nstabilization for networked control systems (NCSs), where the remote controller\nand the local controller operate the linear plant simultaneously. The main\ncontributions are two-fold. Firstly, a necessary and sufficient condition for\nthe finite horizon optimal control problem is given in terms of the two Riccati\nequations. Secondly, it is shown that the system without the additive noise is\nstabilizable in the mean square sense if and only if the two algebraic Riccati\nequations admit the unique solutions, and a sufficient condition is given for\nthe boundedness in the mean square sense of the system with the additive noise.\nNumerical examples about the unmanned aerial vehicle model are shown to\nillustrate the effectiveness of the proposed algorithm.\n", "doi": "10.1016/j.automatica.2018.09.015", "arxiv_id": "1803.01336", "pdf_urls": ["https://arxiv.org/pdf/1803.01336"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 98: 86-94, 2018", "categories": "math.OC"} |
| {"title": "Block Diagonally Dominant Positive Definite Sub-optimal Filters and\n Smoothers", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Kurt S. Riedel"], "abstract": " We examine stochastic dynamical systems where the transition matrix, $\\Phi$,\nand the system noise, $\\bf{\\Gamma}\\bf{Q}\\bf{\\Gamma}^T$, covariance are nearly\nblock diagonal. When $\\bf{H}^T \\bf{R}^{-1} \\bf{H}$ is also nearly block\ndiagonal, where $\\bf{R}$ is the observation noise covariance and $\\bf{H}$ is\nthe observation matrix, our suboptimal filter/smoothers are always positive\nsemi-definite, and have improved numerical properties. Applications for\ndistributed dynamical systems with time dependent pixel imaging are discussed.\n", "doi": null, "arxiv_id": "1803.05083", "pdf_urls": ["https://arxiv.org/pdf/1803.05083"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 29, pp. 779-783 (1993", "categories": "stat.ME cs.SY eess.SY math.RT"} |
| {"title": "On the dynamics of two photons interacting with a two-qubit coherent\n feedback network}", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Guofeng Zhang and Yu Pan"], "abstract": " The purpose of this paper is to study the dynamics of a quantum coherent\nfeedback network composed of two two-level systems (qubits) driven by two\ncounter-propagating photons, one in each input channel. The coherent feedback\nnetwork enhances the nonlinear photon-photon interaction inside the feedback\nloop. By means of quantum stochastic calculus and the input-output framework,\nthe analytic form of the steady-state output two-photon state is derived. Based\non the analytic form, the applications on the Hong-Ou-Mandel (HOM)\ninterferometer and marginally stable single-photon devices using this coherent\nfeedback structure have been demonstrated. The difference between\ncontinuous-mode and single-mode few-photon states is demonstrated.\n", "doi": "10.1016/j.automatica.2020.108978", "arxiv_id": "1803.05630", "pdf_urls": ["https://arxiv.org/pdf/1803.05630"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 117(108978), 1-13, 2020", "categories": "quant-ph"} |
| {"title": "A Tool for Analysis and Synthesis of Heterogeneous Multi-agent Systems\n under Rank-deficient Coupling", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Jin Gyu Lee", "Hyungbo Shim"], "abstract": " The behavior of heterogeneous multi-agent systems is studied when the\ncoupling matrices are possibly all different and/or singular, that is, its rank\nis less than the system dimension. Rank-deficient coupling allows exchange of\nlimited state information, which is suitable for the study of multi-agent\nsystems under output coupling. We present a coordinate change that transforms\nthe heterogeneous multi-agent system into a singularly perturbed form. The slow\ndynamics is still a reduced-order multi-agent system consisting of a weighted\naverage of the vector fields of all agents, and some sub-dynamics of agents.\nThe weighted average is an emergent dynamics, which we call a blended dynamics.\nBy analyzing or synthesizing the blended dynamics, one can predict or design\nthe behavior of a heterogeneous multi-agent system when the coupling gain is\nsufficiently large. For this result, stability of the blended dynamics is\nrequired. Since stability of the individual agent is not asked, the stability\nof the blended dynamics is the outcome of trading off the stability among the\nagents. It can be seen that, under the stability of the blended dynamics, the\ninitial conditions of the individual agents are forgotten as time goes on, and\nthus, the behavior of the synthesized multi-agent system is initialization-free\nand is suitable for plug-and-play operation. As a showcase, we apply the\nproposed tool to four application problems; distributed state estimation for\nlinear systems, practical synchronization of heterogeneous Van der Pol\noscillators, estimation of the number of nodes in a network, and a problem of\ndistributed optimization.\n", "doi": "10.1016/j.automatica.2020.108952", "arxiv_id": "1804.00638", "pdf_urls": ["https://arxiv.org/pdf/1804.00638"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 117, pp. 108952, July 2020", "categories": "cs.SY cs.MA"} |
| {"title": "Structure Discrimination in Block-Oriented Models Using Linear\n Approximations: A Theoretic Framework", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Johan Schoukens", "Rik Pintelon", "Yves Rolain", "Maarten Schoukens", "Koen\n Tiels", "Laurent Vanbeylen", "Anne Van Mulders", "Gerd Vandersteen"], "abstract": " In this paper we show that it is possible to retrieve structural information\nabout complex block-oriented nonlinear systems, starting from linear\napproximations of the nonlinear system around different setpoints.The key idea\nis to monitor the movements of the poles and zeros of the linearized models and\nto reduce the number of candidate models on the basis of these observations.\nBesides the well known open loop single branch Wiener-, Hammerstein-, and\nWiener-Hammerstein systems, we also cover a number of more general structures\nlike parallel (multi branch) Wiener-Hammerstein models, and closed loop block\noriented models, including linear fractional representation (LFR) models.\n", "doi": "10.1016/j.automatica.2014.12.045", "arxiv_id": "1804.09648", "pdf_urls": ["https://arxiv.org/pdf/1804.09648"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 53 (2015) 225-234", "categories": "cs.SY"} |
| {"title": "Regularized Nonparametric Volterra Kernel Estimation", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Georgios Birpoutsoukis", "Anna Marconato", "John Lataire", "and Johan\n Schoukens"], "abstract": " In this paper, the regularization approach introduced recently for\nnonparametric estimation of linear systems is extended to the estimation of\nnonlinear systems modelled as Volterra series. The kernels of order higher than\none, representing higher dimensional impulse responses in the series, are\nconsidered to be realizations of multidimensional Gaussian processes. Based on\nthis, prior information about the structure of the Volterra kernel is\nintroduced via an appropriate penalization term in the least squares cost\nfunction. It is shown that the proposed method is able to deliver accurate\nestimates of the Volterra kernels even in the case of a small amount of data\npoints.\n", "doi": "10.1016/j.automatica.2017.04.014", "arxiv_id": "1804.10435", "pdf_urls": ["https://arxiv.org/pdf/1804.10435"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Birpoutsoukis, G., Marconato, A., Lataire, J., \\& Schoukens, J.\n (2017). Regularized nonparametric Volterra kernel estimation. Automatica, 82,\n 324-327", "categories": "cs.SY"} |
| {"title": "Resilient consensus for multi-agent systems subject to differential\n privacy requirements", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Davide Fiore and Giovanni Russo"], "abstract": " We consider multi-agent systems interacting over directed network topologies\nwhere a subset of agents is adversary/faulty and where the non-faulty agents\nhave the goal of reaching consensus, while fulfilling a differential privacy\nrequirement on their initial conditions. To address this problem, we develop an\nupdate law for the non-faulty agents. Specifically, we propose a modification\nof the so-called Mean-Subsequence-Reduced (MSR) algorithm, the Differentially\nPrivate MSR (DP-MSR) algorithm, and characterize three important properties of\nthe algorithm: correctness, accuracy and differential privacy. We show that if\nthe network topology is $(2f +1)$-robust, then the algorithm allows the\nnon-faulty agents to reach consensus despite the presence of up to $f$ faulty\nagents and we characterize the accuracy of the algorithm. Furthermore, we also\nshow in two important cases that our distributed algorithm can be tuned to\nguarantees differential privacy of the initial conditions and the differential\nprivacy requirement is related to the maximum network degree. The results are\nillustrated via simulations.\n", "doi": "10.1016/j.automatica.2019.04.029", "arxiv_id": "1805.08458", "pdf_urls": ["https://arxiv.org/pdf/1805.08458"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 106, pp. 18-26, 2019", "categories": "cs.SY"} |
| {"title": "Towards Manipulability of Interactive Lagrangian Systems", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Hanlei Wang"], "abstract": " This paper investigates manipulability of interactive Lagrangian systems with\nparametric uncertainty and communication/sensing constraints. Two standard\nexamples are teleoperation with a master-slave system and teaching operation of\nrobots. We here systematically formulate the concept of infinite manipulability\nfor general dynamical systems, and investigate how such a unified motivation\nyields a design paradigm towards guaranteeing the infinite manipulability of\ninteractive dynamical systems and in particular facilitates the design and\nanalysis of nonlinear adaptive controllers for interactive Lagrangian systems.\nSpecifically, based on a new class of dynamic feedback, we propose adaptive\ncontrollers that achieve both the infinite manipulability of the controlled\nLagrangian systems and the robustness with respect to the communication/sensing\nconstraints, mainly owing to the resultant dynamic-cascade framework. The\nproposed paradigm yields the desirable balance between network coupling\nrequirements and controlled dynamics of human-system interaction. We also show\nthat a special case of our main result resolves the longstanding nonlinear\nbilateral teleoperation problem with arbitrary unknown time-varying delay.\nSimulation results show the performance of the interactive robotic systems\nunder the proposed adaptive controllers.\n", "doi": "10.1016/j.automatica.2020.108913", "arxiv_id": "1806.06723", "pdf_urls": ["https://arxiv.org/pdf/1806.06723"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 119: 108913, 2020", "categories": "cs.SY cs.RO math.OC"} |
| {"title": "An Influence Network Model to Study Discrepancies in Expressed and\n Private Opinions", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Mengbin Ye", "Yuzhen Qin", "Alain Govaert", "Brian D.O. Anderson", "Ming Cao"], "abstract": " In many social situations, a discrepancy arises between an individual's\nprivate and expressed opinions on a given topic. Motivated by Solomon Asch's\nseminal experiments on social conformity and other related socio-psychological\nworks, we propose a novel opinion dynamics model to study how such a\ndiscrepancy can arise in general social networks of interpersonal influence.\nEach individual in the network has both a private and an expressed opinion: an\nindividual's private opinion evolves under social influence from the expressed\nopinions of the individual's neighbours, while the individual determines his or\nher expressed opinion under a pressure to conform to the average expressed\nopinion of his or her neighbours, termed the local public opinion. General\nconditions on the network that guarantee exponentially fast convergence of the\nopinions to a limit are obtained. Further analysis of the limit yields several\nsemi-quantitative conclusions, which have insightful social interpretations,\nincluding the establishing of conditions that ensure every individual in the\nnetwork has such a discrepancy. Last, we show the generality and validity of\nthe model by using it to explain and predict the results of Solomon Asch's\nseminal experiments.\ndoi10.1016/j.automatica.2019.05.059arxiv_id1806.11236pdf_urlshttps://arxiv.org/pdf/1806.11236githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 107(9): pp. 371-381, Sept 2019categoriescs.SI cs.MA cs.SY eess.SY math.DS |
| titleOn the Ergodic Control of Ensemblesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAndre R. FioravantiJakub MarecekRobert N. ShortenMatheus SouzaFabian R. Wirthabstract Across smart-grid and smart-city application domains, there are many problems\nwhere an ensemble of agents is to be controlled such that both the aggregate\nbehaviour and individual-level perception of the system's performance are\nacceptable. In many applications, traditional PI control is used to regulate\naggregate ensemble performance. Our principal contribution in this note is to\ndemonstrate that PI control may not be always suitable for this purpose, and in\nsome situations may lead to a loss of ergodicity for closed-loop systems.\nBuilding on this observation, a theoretical framework is proposed to both\nanalyse and design control systems for the regulation of large scale ensembles\nof agents with a probabilistic intent. Examples are given to illustrate our\nresults.\n", "doi": "10.1016/j.automatica.2019.06.035", "arxiv_id": "1807.03256", "pdf_urls": ["https://arxiv.org/pdf/1807.03256"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 108, October 2019", "categories": "math.OC cs.SY"} |
| {"title": "Decimation analysis in the signal processing of current to detect broken\n bars in induction machine", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["J. S. Moreira", "P. C. M. Lamim Filho", "L. M. R. Baccarini", "E. G.\n Nepomuceno", "P. F. S. Guedes"], "abstract": " This paper presents a study on the reduction of the sampling frequency of the\ncurrent signals of an induction motor, the reductions are performed by means\ntime-decimation technique for digital signal processing. We have used the Fast\nFourier Transform to obtain the fault signal spectrum of broken bars. The\nresults have shown how the decimation technique significantly reduces the\nnumber of operations and the time required to calculate the Fast Fourier\nTransform without loss of information. This approach provides a better\nperformance of embedded systems for fault diagnosis based on characteristic of\namplitude signal modulation.\n", "doi": null, "arxiv_id": "1807.10930", "pdf_urls": ["https://arxiv.org/pdf/1807.10930"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "CBA 2018 - XXII Congresso Brasileiro de Automatica, Joao Pessoa,\n p. 1-6", "categories": "eess.SP"} |
| {"title": "Deep Reinforcement Learning for Wireless Sensor Scheduling in\n Cyber-Physical Systems", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Alex S. Leong", "Arunselvan Ramaswamy", "Daniel E. Quevedo", "Holger Karl", "Ling Shi"], "abstract": " In many Cyber-Physical Systems, we encounter the problem of remote state\nestimation of geographically distributed and remote physical processes. This\npaper studies the scheduling of sensor transmissions to estimate the states of\nmultiple remote, dynamic processes. Information from the different sensors have\nto be transmitted to a central gateway over a wireless network for monitoring\npurposes, where typically fewer wireless channels are available than there are\nprocesses to be monitored. For effective estimation at the gateway, the sensors\nneed to be scheduled appropriately, i.e., at each time instant one needs to\ndecide which sensors have network access and which ones do not. To address this\nscheduling problem, we formulate an associated Markov decision process (MDP).\nThis MDP is then solved using a Deep Q-Network, a recent deep reinforcement\nlearning algorithm that is at once scalable and model-free. We compare our\nscheduling algorithm to popular scheduling algorithms such as round-robin and\nreduced-waiting-time, among others. Our algorithm is shown to significantly\noutperform these algorithms for many example scenarios.\n", "doi": null, "arxiv_id": "1809.05149", "pdf_urls": ["https://arxiv.org/pdf/1809.05149"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 113, 2020", "categories": "cs.SY"} |
| {"title": "A unitary distributed subgradient method for multi-agent optimization\n with different coupling sources", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Changxin Liu", "Huiping Li", "Yang Shi"], "abstract": " In this work, we first consider distributed convex constrained optimization\nproblems where the objective function is encoded by multiple local and possibly\nnonsmooth objectives privately held by a group of agents, and propose a\ndistributed subgradient method with double averaging (abbreviated as ${\\rm\nDSA_2}$) that only requires peer-to-peer communication and local computation to\nsolve the global problem. The algorithmic framework builds on dual methods and\ndynamic average consensus; the sequence of test points is formed by iteratively\nminimizing a local dual model of the overall objective where the coefficients,\ni.e., approximated subgradients of the objective, are supplied by the dynamic\naverage consensus scheme. We theoretically show that ${\\rm DSA_2}$ enjoys\nnon-ergodic convergence properties, i.e., the local minimizing sequence itself\nis convergent, a distinct feature that cannot be found in existing results.\nSpecifically, we establish a convergence rate of $O(\\frac{1}{\\sqrt{t}})$ in\nterms of objective function error. Then, extensions are made to tackle\ndistributed optimization problems with coupled functional constraints by\ncombining ${\\rm DSA_2}$ and dual decomposition. This is made possible by\nLagrangian relaxation that transforms the coupling in constraints of the primal\nproblem into that in cost functions of the dual, thus allowing us to solve the\ndual problem via ${\\rm DSA_2}$. Both the dual objective error and the quadratic\npenalty for the coupled constraint are proved to converge at a rate of\n$O(\\frac{1}{\\sqrt{t}})$, and the primal objective error asymptotically\nvanishes. Numerical experiments and comparisons are conducted to illustrate the\nadvantage of the proposed algorithms and validate our theoretical findings.\n", "doi": "10.1016/j.automatica.2020.108834", "arxiv_id": "1809.06008", "pdf_urls": ["https://arxiv.org/pdf/1809.06008"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 114, pp. 1-12, 2020", "categories": "math.OC"} |
| {"title": "Retrofit Control with Approximate Environment Modeling", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Takayuki Ishizaki", "Takahiro Kawaguchi", "Hampei Sasahara", "and Jun-ichi\n Imura"], "abstract": " In this paper, we develop a retrofit control method with approximate\nenvironment modeling. Retrofit control is a modular control approach for a\ngeneral stable network system whose subsystems are supposed to be managed by\ntheir corresponding subsystem operators. From the standpoint of a single\nsubsystem operator who performs the design of a retrofit controller, the\nsubsystems managed by all other operators can be regarded as an environment,\nthe complete system model of which is assumed not to be available. The proposed\nretrofit control with approximate environment modeling has an advantage that\nthe stability of the resultant control system is robustly assured regardless of\nnot only the stability of approximate environment models, but also the\nmagnitude of modeling errors, provided that the network system before\nimplementing retrofit control is originally stable. This robustness property is\npractically significant to incorporate existing identification methods of\nunknown environments, because the accuracy of identified models may neither be\nreliable nor assurable in reality. Furthermore, we conduct a control\nperformance analysis to show that the resultant performance can be regulated by\nadjusting the accuracy of approximate environment modeling. The efficiency of\nthe proposed retrofit control is shown by numerical experiments on a network of\nsecond-order oscillators.\n", "doi": "10.1016/j.automatica.2019.06.010", "arxiv_id": "1810.01141", "pdf_urls": ["https://arxiv.org/pdf/1810.01141"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 107, pp.442-453, 2019", "categories": "cs.SY math.OC"} |
| {"title": "Distributed Conditional Cooperation Model Predictive Control of\n Interconnected Microgrids", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["A. K. Sampathirao", "S. Hofmann", "J. Raisch", "and C. A. Hans"], "abstract": " In this paper, we propose a model predictive control based operation strategy\nthat allows for power exchange between interconnected microgrids. Particularly,\nthe approach ensures that each microgrid benefits from power exchange with\nothers. This is realised by including a condition which is based on the\nislanded operation cost. The overall model predictive control problem is posed\nas a mixed-integer quadratically-constrained program and solved using a\ndistributed algorithm that iteratively updates continuous and integer\nvariables. For this algorithm, termination, feasibility and computational\nproperties are discussed. The performance and the computational benefits of the\nproposed strategy are highlighted in an illustrative case study.\n", "doi": "10.1016/j.automatica.2023.111258", "arxiv_id": "1810.03361", "pdf_urls": ["https://arxiv.org/pdf/1810.03361"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 157, p. 111258, 2023", "categories": "cs.SY"} |
| {"title": "ISS Property with Respect to Boundary Disturbances for a Class of\n Riesz-Spectral Boundary Control Systems", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Hugo Lhachemi and Robert Shorten"], "abstract": " This paper deals with the establishment of Input-to-State Stability (ISS)\nestimates for infinite dimensional systems with respect to both boundary and\ndistributed disturbances. First, a new approach is developed for the\nestablishment of ISS estimates for a class of Riesz-spectral boundary control\nsystems satisfying certain eigenvalue constraints. Second, a concept of weak\nsolutions is introduced in order to relax the disturbances regularity\nassumptions required to ensure the existence of classical solutions. The\nproposed concept of weak solutions, that applies to a large class of boundary\ncontrol systems which is not limited to the Riesz-spectral ones, provides a\nnatural extension of the concept of both classical and mild solutions. Assuming\nthat an ISS estimate holds true for classical solutions, we show the existence,\nthe uniqueness, and the ISS property of the weak solutions.\n", "doi": "10.1016/j.automatica.2019.108504", "arxiv_id": "1810.03553", "pdf_urls": ["https://arxiv.org/pdf/1810.03553"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 109, November 2019, 108504", "categories": "math.OC cs.SY"} |
| {"title": "A Transformation-Proximal Bundle Algorithm for Multistage Adaptive\n Robust Optimization and Application to Constrained Robust Optimal Control", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Chao Ning", "Fengqi You"], "abstract": " This paper presents a novel transformation-proximal bundle algorithm for\nmultistage adaptive robust optimization problems. By partitioning recourse\ndecisions into state and control decisions, the proposed algorithm applies\naffine control policy only to state decisions and allows control decisions to\nbe fully adaptive, thus transforming the original problem into an equivalent\ntwo-stage Adaptive Robust Optimization (ARO) problem. Importantly, this\nmulti-to-two transformation is general enough to be employed with any two-stage\nARO solution algorithms, thus opening a new avenue for a variety of multistage\nARO algorithms. The proximal bundle method is developed for the resulting\ntwo-stage problem along with convergence analysis. In an inventory control\napplication, the affine disturbance-feedback control policy suffers from a\nsevere suboptimality with an average gap of 34.88%, while the proposed\nalgorithm generates an average gap of merely 1.68%.\n", "doi": "10.1016/j.automatica.2019.108802", "arxiv_id": "1810.05931", "pdf_urls": ["https://arxiv.org/pdf/1810.05931"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 113, March 2020, 108802", "categories": "math.OC cs.SY"} |
| {"title": "Generalized Lyapunov criteria on finite-time stability of stochastic\n nonlinear systems", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Xin Yu and Juliang Yin and Suiyang Khoo"], "abstract": " This paper considers the problem of finite-time stability for stochastic\nnonlinear systems. A new Lyapunov theorem of stochastic finite-time stability\nis proposed, and an important corollary is obtained. Some comparisons with the\nexisting results are given, and it shows that this new Lyapunov theorem not\nonly is a generalization of classical stochastic finite-time theorem, but also\nreveals the important role of white-noise in finite-time stabilizing stochastic\nsystems. In addition, multiple Lyapunov functions-based criteria on stochastic\nfinite-time stability are presented, which further relax the constraint of the\ninfinitesimal generator $\\mathcal{L}V$. Some examples are constructed to show\nsignificant features of the proposed theorems. Finally, simulation results are\npresented to demonstrate the theoretical analysis.\n", "doi": "10.1016/j.automatica.2019.05.048", "arxiv_id": "1810.07927", "pdf_urls": ["https://arxiv.org/pdf/1810.07927"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2019", "categories": "math.PR"} |
| {"title": "Approximation of nonnegative systems by moving averages of fixed order", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Lorenzo Finesso", "Peter Spreij"], "abstract": " We pose the approximation problem for scalar nonnegative input-output systems\nvia impulse response convolutions of finite order, i.e. finite order moving\naverages, based on repeated observations of input/output signal pairs. The\nproblem is converted into a nonnegative matrix factorization with special\nstructure for which we use Csisz\\'ar's I-divergence as the criterion of\noptimality. Conditions are given, on the input/output data, that guarantee the\nexistence and uniqueness of the minimum. We propose an algorithm of the\nalternating minimization type for I-divergence minimization, and present its\nasymptotic behavior. For the case of noisy observations we give the large\nsample properties of the statistical version of the minimization problem for\ndifferent observation regimes. Numerical experiments confirm the asymptotic\nresults and exhibit fast convergence of the proposed algorithm.\n", "doi": "10.1016/j.automatica.2019.05.007", "arxiv_id": "1810.10135", "pdf_urls": ["https://arxiv.org/pdf/1810.10135"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 107, pp 1-8, 2019", "categories": "math.OC"} |
| {"title": "Passivity-Based Analysis of Sampled and Quantized Control\n Implementations", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Xiangru Xu", "Necmiye Ozay", "Vijay Gupta"], "abstract": " This paper studies the performance of a continuous controller when\nimplemented on digital devices via sampling and quantization, by leveraging\npassivity analysis. Degradation of passivity indices from a continuous-time\ncontrol system to its sampled, input and output quantized model is studied\nusing a notion of quasi-passivity. Based on that, the passivity property of a\nfeedback-connected system where the continuous controller is replaced by its\nsampled and quantized model is studied, and conditions that ensure the state\nboundedness of the interconnected system are provided. Additionally, the\napproximate bisimulation-based control implementation where the controller is\nreplaced by its approximate bisimilar symbolic model whose states are also\nquantized is analyzed. Several examples are provided to illustrate the\ntheoretical results.\n", "doi": "10.1016/j.automatica.2020.109064", "arxiv_id": "1810.11950", "pdf_urls": ["https://arxiv.org/pdf/1810.11950"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2020", "categories": "math.OC cs.SY"} |
| {"title": "Automatically Explaining Machine Learning Prediction Results: A\n Demonstration on Type 2 Diabetes Risk Prediction", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Gang Luo"], "abstract": " Background: Predictive modeling is a key component of solutions to many\nhealthcare problems. Among all predictive modeling approaches, machine learning\nmethods often achieve the highest prediction accuracy, but suffer from a\nlong-standing open problem precluding their widespread use in healthcare. Most\nmachine learning models give no explanation for their prediction results,\nwhereas interpretability is essential for a predictive model to be adopted in\ntypical healthcare settings. Methods: This paper presents the first complete\nmethod for automatically explaining results for any machine learning predictive\nmodel without degrading accuracy. We did a computer coding implementation of\nthe method. Using the electronic medical record data set from the Practice\nFusion diabetes classification competition containing patient records from all\n50 states in the United States, we demonstrated the method on predicting type 2\ndiabetes diagnosis within the next year. Results: For the champion machine\nlearning model of the competition, our method explained prediction results for\n87.4% of patients who were correctly predicted by the model to have type 2\ndiabetes diagnosis within the next year. Conclusions: Our demonstration showed\nthe feasibility of automatically explaining results for any machine learning\npredictive model without degrading accuracy.\n", "doi": "10.1186/s13755-016-0015-4", "arxiv_id": "1812.02852", "pdf_urls": ["https://arxiv.org/pdf/1812.02852"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Gang Luo. Automatically Explaining Machine Learning Prediction\n Results: A Demonstration on Type 2 Diabetes Risk Prediction. Health\n Information Science and Systems, Vol. 4, No. 2, Mar. 2016", "categories": "cs.LG stat.ML"} |
| {"title": "Online learning with stability guarantees: A memory-based real-time\n model predictive controller", "year": 2018, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Lukas Schwenkel", "Meriem Gharbi", "Sebastian Trimpe", "Christian Ebenbauer"], "abstract": " We propose and analyze a real-time model predictive control (MPC) scheme that\nutilizes stored data to improve its performance by learning the value function\nonline with stability guarantees. For linear and nonlinear systems, a learning\nmethod is presented that makes use of basic analytic properties of the cost\nfunction and is proven to learn the MPC control law and the value function on\nthe limit set of the closed-loop state trajectory. The main idea is to generate\na smart warm start based on historical data that improves future data points\nand thus future warm starts. We show that these warm starts are asymptotically\nexact and converge to the solution of the MPC optimization problem. Thereby,\nthe suboptimality of the applied control input resulting from the real-time\nrequirements vanishes over time. Simulative examples show that existing\nreal-time MPC schemes can be improved by storing data and the proposed learning\nscheme.\n", "doi": "10.1016/j.automatica.2020.109247", "arxiv_id": "1812.09582", "pdf_urls": ["https://arxiv.org/pdf/1812.09582"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 122, 109247, 2020", "categories": "math.OC"} |
| {"title": "FBstab: A Stabilized Semismooth Quadratic Programming Algorithm with\n Applications in Model Predictive Control", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Dominic Liao-McPherson", "Ilya Kolmanovsky"], "abstract": " This paper introduces the proximally stabilized Fischer-Burmeister method\n(FBstab); a new algorithm for convex quadratic programming that synergistically\ncombines the proximal point algorithm with a primal-dual semismooth Newton-type\nmethod. FBstab is numerically robust, easy to warmstart, handles degenerate\nprimal-dual solutions, detects infeasibility/unboundedness and requires only\nthat the Hessian matrix be positive semidefinite. We outline the algorithm,\nprovide convergence and convergence rate proofs, report some numerical results\nfrom model predictive control benchmarks, and also include experimental\nresults. We show that FBstab is competitive with and often superior to, state\nof the art methods, has attractive scaling properties, and is especially\npromising for model predictive control applications.\n", "doi": "10.1016/j.automatica.2019.108801", "arxiv_id": "1901.04046", "pdf_urls": ["https://arxiv.org/pdf/1901.04046"], "github": ["https://github.com/dliaomcp/fbstab", "https://github.com/dliaomcp/fbstab-matlab"], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 113, p. 108801, 2020", "categories": "math.OC"} |
| {"title": "Funnel control for a moving water tank", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Thomas Berger", "Marc Puche and Felix Schwenninger"], "abstract": " We study tracking control for a nonlinear moving water tank system modelled\nby the linearized Saint-Venant equations, where the output is given by the\nposition of the tank and the control input is the force acting on it. For a\ngiven reference signal, the objective is that the tracking error evolves within\na pre-specified performance funnel. Exploiting recent results in funnel\ncontrol, this can be achieved by showing that inter alia the system's internal\ndynamics are bounded-input, bounded-output stable.\ndoi10.1016/j.automatica.2021.109999arxiv_id1902.00586pdf_urlshttps://arxiv.org/pdf/1902.00586githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 135, January 2022, 109999categoriesmath.OC |
| titleA Generalization of Linear Positive Systems with Applications to\n Nonlinear Systems: Invariant Sets and the Poincar\\'{e}-Bendixson Property", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Eyal Weiss", "Michael Margaliot"], "abstract": " The dynamics of linear positive systems map the positive orthant to itself.\nIn other words, it maps a set of vectors with zero sign variations to itself.\nThis raises the following question: what linear systems map the set of vectors\nwith $k$ sign variations to itself? We address this question using tools from\nthe theory of cooperative dynamical systems and the theory of totally positive\nmatrices. This yields a generalization of positive linear systems called\n$k$-positive linear systems, that reduces to positive systems for $k=1$. We\ndescribe applications of this new type of systems to the analysis of nonlinear\ndynamical systems. In particular, we show that such systems admit certain\nexplicit invariant sets, and for the case $k=2$ establish the\nPoincar\\'{e}-Bendixson property for any bounded trajectory.\ndoi10.1016/j.automatica.2020.109358arxiv_id1902.01630pdf_urlshttps://arxiv.org/pdf/1902.01630githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 123 (2021): 109358categoriescs.SY math.DS |
| titleAn LMI Condition for the Robustness of Constant-Delay Linear Predictor\n Feedback with Respect to Uncertain Time-Varying Input Delaysyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsHugo LhachemiChristophe Prieur and Robert Shortenabstract This paper discusses the robustness of the constant-delay predictor feedback\nin the case of an uncertain time-varying input delay. Specifically, we study\nthe stability of the closed-loop system when the predictor feedback is designed\nbased on the knowledge of the nominal value of the time-varying delay. By\nresorting to an adequate Lyapunov-Krasovskii functional, we derive an LMI-based\nsufficient condition ensuring the exponential stability of the closed-loop\nsystem for small enough variations of the time-varying delay around its nominal\nvalue. These results are extended to the feedback stabilization of a class of\ndiagonal infinite-dimensional boundary control systems in the presence of a\ntime-varying delay in the boundary control input.\ndoi10.1016/j.automatica.2019.108551arxiv_id1902.07778pdf_urlshttps://arxiv.org/pdf/1902.07778githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 109, November 2019, 108551categoriesmath.OC cs.SY |
| titleTime Distributed Optimization for Model Predictive Control: Stability,\n Robustness, and Constraint Satisfactionyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsDominic Liao-McPhersonMarco Nicotraand Ilya Kolmanovskyabstract Time distributed optimization is an implementation strategy that can\nsignificantly reduce the computational burden of model predictive control by\nexploiting its robustness to incomplete optimization. When using this strategy,\noptimization iterations are distributed over time by maintaining a running\nsolution estimate for the optimal control problem and updating it at each\nsampling instant. The resulting controller can be viewed as a dynamic\ncompensator which is placed in closed-loop with the plant. This paper presents\na general systems theoretic analysis framework for time distributed\noptimization. The coupled plant-optimizer system is analyzed using\ninput-to-state stability concepts and sufficient conditions for stability and\nconstraint satisfaction are derived. When applied to time distributed\nsequential quadratic programming, the framework significantly extends the\nexisting theoretical analysis for the real-time iteration scheme. Numerical\nsimulations are presented that demonstrate the effectiveness of the scheme.\ndoi10.1016/j.automatica.2020.108973arxiv_id1903.02605pdf_urlshttps://arxiv.org/pdf/1903.02605githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 117, p. 108973, 2020categoriesmath.OC |
| titleOrbital stabilization of nonlinear systems via Mexican sombrero energy\n shaping and pumping-and-damping injectionyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsBowen YiRomeo OrtegaDongjun WuWeidong Zhangabstract In this paper we show that a slight modification to the widely popular\ninterconnection and damping assignment passivity-based control\nmethod---originally proposed for stabilization of equilibria of nonlinear\nsystems---allows us to provide a solution to the more challenging orbital\nstabilization problem. Two different, though related, ways how this procedure\ncan be applied are proposed. First, the assignment of an energy function that\nhas a minimum in a closed curve, i.e., with the shape of a Mexican sombrero.\nSecond, the use of a damping matrix that changes \ according to the\nposition of the state trajectory relative to the desired orbit, that is,\npumping or dissipating energy. The proposed methodologies are illustrated with\nthe example of the induction motor and prove that it yields the industry\nstandard field oriented control.\ndoi10.1016/j.automatica.2019.108661arxiv_id1903.04070pdf_urlshttps://arxiv.org/pdf/1903.04070githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2020categoriescs.SY |
| titleEstimation of Photovoltaic Generation Forecasting Models using Limited\n Informationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsGianni Bianchini and Daniele Pepe and Antonio Vicinoabstract This work deals with the problem of estimating a photovoltaic generation\nforecasting model in scenarios where measurements of meteorological variables\n(i.e. solar irradiance and temperature) at the plant site are not available. A\nnovel algorithm for the estimation of the parameters of the well-known PVUSA\nmodel of a photovoltaic plant is proposed. Such a method is characterized by a\nlow computational complexity, and efficiently exploits only power generation\nmeasurements, a theoretical clear-sky irradiance model, and temperature\nforecasts provided by a meteorological service. An extensive experimental\nvalidation of the proposed method on real data is also presented.\ndoi10.1016/j.automatica.2019.108688arxiv_id1903.04827pdf_urlshttps://arxiv.org/pdf/1903.04827githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2019categoriescs.SY |
| titleRegularity/Controllability/Observability of an NDS with Descriptor Form\n Subsystems and Generalized LFTsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsTong Zhouabstract This paper investigates regularity, controllability and observability for a\nnetworked dynamic system (NDS) with its subsystems being described in a\ndescriptor form and system matrices of each subsystem being represented by a\ngeneralized linear fractional transformation (GLFT) of its parameters. Except a\nwell-posedness condition, no any other constraints are put on either parameters\nor connections of a subsystem. Based on the Kronecker canonical form (KCF) of a\nmatrix pencil, some matrix rank based necessary and sufficient conditions are\nestablished respectively for the regularity and complete\ncontrollability/observability of the NDS, in which the associated matrix\ndepends affinely on both subsystem parameters and subsystem connections. These\nconditions keep the property that all the involved numerical computations are\nperformed on each subsystem independently, which is attractive in the analysis\nand synthesis of a large scale NDS. Moreover, some explicit and easily\ncheckable requirements are derived for subsystem dynamics/parameters with which\na completely controllable/observable NDS can be constructed more easily.\ndoi10.1016/j.automatica.2020.109153arxiv_id1903.06018pdf_urlshttps://arxiv.org/pdf/1903.06018githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 121 (2020) 109153 (free download before September 23,\n 2020, using https://authors.elsevier.com/a/1bWLJ1AMv1YYQ)categoriescs.SY |
| titleRobust matrix commutator conditions for stability of switched linear\n systems under restricted switchingyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAtreyee Kundu and Debasish Chatterjeeabstract This article treats global uniform exponential stability (GUES) of\ndiscrete-time switched linear systems under restricted switching. Given\nadmissible minimum and maximum dwell times, we provide sufficient conditions on\nthe subsystems under which they admit a set of switching signals that obeys the\ngiven restrictions on dwell times and preserves stability of the resulting\nswitched system. Our analysis relies on combinatorial arguments applied to\nmatrix commutators and avoids the employment of Lyapunov-like functions. The\nproposed set of stabilizing switching signals is characterized in terms of\nduration of activation of Schur stable subsystems and non-consecutive\nactivation of distinct unstable subsystems.\ndoi10.1016/j.automatica.2020.108904arxiv_id1903.10249pdf_urlshttps://arxiv.org/pdf/1903.10249githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 115, 2020, article no.: 108904categoriescs.SY math.OC |
| titleDetectability and Observer Design for Switched Differential Algebraic\n Equationsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAneel Tanwani and Stephan Trennabstract This paper studies detectability for switched linear differential-algebraic\nequations (DAEs) and its application to the synthesis of observers, which\ngenerate asymptotically converging state estimates. Equating detectability to\nasymptotic stability of zero-output-constrained state trajectories, and\nbuilding on our work on interval-wise observability, we propose the notion of\ninterval-wise detectability: If the output of the system is constrained to be\nidentically zero over an interval, then the norm of the corresponding state\ntrajectories scales down by a certain factor at the end of that interval.\nConditions are provided under which the interval-wise detectability leads to\nasymptotic stability of zero-output-constrained state trajectories. An\napplication is demonstrated in designing state estimators. Decomposing the\nstate into observable and unobservable components, we show that if the\nobservable component of the system is reset appropriately and persistently,\nthen the estimation error converges to zero asymptotically under the\ninterval-wise detectability assumption.\ndoi10.1016/j.automatica.2018.10.043arxiv_id1904.00124pdf_urlshttps://arxiv.org/pdf/1904.00124githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 99, Pages 289-300, 2019categoriesmath.OC cs.SY math.DS |
| titleISS Lyapunov Functions for Cascade Switched Systems and Sampled-Data\n Controlyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsGuangXue Zhang and Aneel Tanwaniabstract Input-to-state stability (ISS) of switched systems is studied where the\nindividual subsystems are connected in a serial cascade configuration, and the\nstates are allowed to reset at switching times. An ISS Lyapunov function is\nassociated to each of the two blocks connected in cascade, and these functions\nare used as building blocks for constructing ISS Lyapunov function for the\ninterconnected system. The derivative of individual Lyapunov functions may be\nbounded by nonlinear decay functions, and the growth in the value of Lyapunov\nfunction at switching times may also be a nonlinear function of the value of\nother Lyapunov functions. The stability of overall hybrid system is analyzed by\nconstructing a newly constructed ISS-Lyapunov function and deriving lower\nbounds on the average dwell-time. The particular case of linear subsystems and\nquadratic Lyapunov functions is also studied. The tools are also used for\nstudying the observer-based feedback stabilization of a nonlinear switched\nsystem with event-based sampling of the output and control inputs. We design\ndynamic sampling algorithms based on the proposed Lyapunov functions and\nanalyze the stability of the resulting closed-loop system.\ndoi10.1016/j.automatica.2019.03.028arxiv_id1904.00616pdf_urlshttps://arxiv.org/pdf/1904.00616githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 105:216--227, July 2019categoriescs.SY math.DS math.OC |
| titleOn Maximal Robust Positively Invariant Sets in Constrained Nonlinear\n Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsWillem EsterhuizenTim AschenbruckStefan Streifabstract In this technical communique we study the maximal robust positively invariant\nset for state-constrained continuous-time nonlinear systems subjected to a\nbounded disturbance. Extending results from the theory of barriers, we show\nthat this set is closed and that its boundary consists of two complementary\nparts, one of which we name the invariance barrier, which consists of\ntrajectories that satisfy the maximum principle.\ndoi10.1016/j.automatica.2020.109044arxiv_id1904.01985pdf_urlshttps://arxiv.org/pdf/1904.01985githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 119, 109044 (2020)categoriesmath.OC cs.SY eess.SY |
| titleA system-theoretic framework for privacy preservation in continuous-time\n multiagent dynamicsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsClaudio Altafiniabstract In multiagent dynamical systems, privacy protection corresponds to avoid\ndisclosing the initial states of the agents while accomplishing a distributed\ntask. The system-theoretic framework described in this paper for this scope,\ndenoted dynamical privacy, relies on introducing output maps which act as\nmasks, rendering the internal states of an agent indiscernible by the other\nagents as well as by external agents monitoring all communications. Our output\nmasks are local (i.e., decided independently by each agent), time-varying\nfunctions asymptotically converging to the true states. The resulting masked\nsystem is also time-varying, and has the original unmasked system as its limit\nsystem. When the unmasked system has a globally exponentially stable\nequilibrium point, it is shown in the paper that the masked system has the same\npoint as a global attractor. It is also shown that existence of equilibrium\npoints in the masked system is not compatible with dynamical privacy.\nApplication of dynamical privacy to popular examples of multiagent dynamics,\nsuch as models of social opinions, average consensus and synchronization, is\ninvestigated in detail.\ndoi10.1016/j.automatica.2020.109253arxiv_id1904.11246pdf_urlshttps://arxiv.org/pdf/1904.11246githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 122, 2020, 109253categoriescs.SY math.OC |
| titlePersonalized Optimization with User's Feedback", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Andrea Simonetto", "Emiliano Dall'AneseJulien Monteiland Andrey\n Bernsteinabstract This paper develops an online algorithm to solve a time-varying optimization\nproblem with an objective that comprises a known time-varying cost and an\nunknown function. This problem structure arises in a number of engineering\nsystems and cyber-physical systems where the known function captures\ntime-varying engineering costs, and the unknown function models user's\nsatisfaction; in this context, the objective is to strike a balance between\ngiven performance metrics and user's satisfaction. Key challenges related to\nthe problem at hand are related to (1) the time variability of the problem, and\n(2) the fact that learning of the user's utility function is performed\nconcurrently with the execution of the online algorithm. This paper leverages\nGaussian processes (GP) to learn the unknown cost function from noisy\nfunctional evaluation and build pertinent upper confidence bounds. Using the GP\nformalism, the paper then advocates time-varying optimization tools to design\nan online algorithm that exhibits tracking of the oracle-based optimal\ntrajectory within an error ball, while learning the user's satisfaction\nfunction with no-regret. The algorithmic steps are inexact, to account for\npossible limited computational budgets or real-time implementation\nconsiderations. Numerical examples are illustrated based on a problem related\nto vehicle platooning.\ndoi10.1016/j.automatica.2021.109767arxiv_id1905.00775pdf_urlshttps://arxiv.org/pdf/1905.00775githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 131, 2021categoriesmath.OC cs.SY |
| titleOn the smoothness of nonlinear system identificationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAnt\\^onio H. RibeiroKoen TielsJack UmenbergerThomas B. Sch\\\, ], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2019, : , : , : [], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2019, : , : , : [, , ], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2019, : , : , : [], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2019, : , : , : [], : , : , : , : [], : [], : true, : [], : , : , : } |
| {: , : 2019, : , : , : [, oberlJohannes Diwoldabstract We prove that every flat nonlinear discrete-time system can be decomposed by\ncoordinate transformations into a smaller-dimensional subsystem and an\nendogenous dynamic feedback. For flat continuous-time systems, no comparable\nresult is available. The advantage of such a decomposition is that the complete\nsystem is flat if and only if the subsystem is flat. Thus, by repeating the\ndecomposition at most $n-1$ times, where $n$ is the dimension of the state\nspace, the flatness of a discrete-time system can be checked in an algorithmic\nway. If the system is flat, then the algorithm yields a flat output which only\ndepends on the state variables. Hence, every flat discrete-time system has a\nflat output which does not depend on the inputs and their forward-shifts.\nAgain, no comparable result for flat continuous-time systems is available. The\nalgorithm requires in each decomposition step the construction of state- and\ninput transformations, which are obtained by straightening out certain vector\nfields or distributions with the flow-box theorem or the Frobenius theorem.\nThus, from a computational point of view, only the calculation of flows and the\nsolution of algebraic equations is needed. We illustrate our results by two\nexamples.\ndoi10.1016/j.automatica.2021.109828arxiv_id1907.00596pdf_urlshttps://arxiv.org/pdf/1907.00596githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 2021categoriesmath.OC math.DG math.DS |
| titleA stochastic MPC scheme for distributed systems with multiplicative\n uncertaintyyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsChristoph MarkSteven Liuabstract This paper presents a Distributed Stochastic Model Predictive Control\nalgorithm for networks of linear systems with multiplicative uncertainties and\nlocal chance constraints on the states and control inputs. The chance\nconstraints are approximated via Cantelli's inequality by means of expected\nvalue and covariance. The cooperative control algorithm is based on the\ndistributed Alternating Direction Method of Multipliers, which renders the\ncontroller fully distributedly implementable, recursively feasible and ensures\npoint-wise convergence of the states. The aforementioned properties are\nguaranteed through a properly selected distributed invariant set and\ndistributed terminal constraints for the mean and covariance. The paper closes\nwith an example highlighting the chance constraint satisfaction, numerical\nproperties and scalability of our approach.\n", "doi": "10.1016/j.automatica.2022.110208", "arxiv_id": "1908.09337", "pdf_urls": ["https://arxiv.org/pdf/1908.09337"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 140 (2022) 110208", "categories": "math.OC"} |
| {"title": "Constraint Learning for Control Tasks with Limited Duration Barrier\n Functions", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Motoya Ohnishi", "Gennaro Notomista", "Masashi Sugiyama", "Magnus Egerstedt"], "abstract": " When deploying autonomous agents in unstructured environments over sustained\nperiods of time, adaptability and robustness oftentimes outweigh optimality as\na primary consideration. In other words, safety and survivability constraints\nplay a key role and in this paper, we present a novel, constraint-learning\nframework for control tasks built on the idea of constraints-driven control.\nHowever, since control policies that keep a dynamical agent within state\nconstraints over infinite horizons are not always available, this work instead\nconsiders constraints that can be satisfied over some finite time horizon T >\n0, which we refer to as limited-duration safety. Consequently, value function\nlearning can be used as a tool to help us find limited-duration safe policies.\nWe show that, in some applications, the existence of limited-duration safe\npolicies is actually sufficient for long-duration autonomy. This idea is\nillustrated on a swarm of simulated robots that are tasked with covering a\ngiven area, but that sporadically need to abandon this task to charge\nbatteries. We show how the battery-charging behavior naturally emerges as a\nresult of the constraints. Additionally, using a cart-pole simulation\nenvironment, we show how a control policy can be efficiently transferred from\nthe source task, balancing the pole, to the target task, moving the cart to one\ndirection without letting the pole fall down.\n", "doi": "10.1016/j.automatica.2021.109504", "arxiv_id": "1908.09506", "pdf_urls": ["https://arxiv.org/pdf/1908.09506"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 127 (2021)", "categories": "eess.SY cs.SY"} |
| {"title": "Robust Hybrid Zero-Order Optimization Algorithms with Acceleration via\n Averaging in Time", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Jorge I. Poveda and Na Li"], "abstract": " We study novel robust zero-order algorithms with acceleration for the\nsolution of real-time optimization problems. In particular, we propose a family\nof extremum seeking dynamics that can be universally modeled as singularly\nperturbed hybrid dynamical systems with restarting mechanisms. From this family\nof dynamics, we synthesize four fast algorithms for the solution of convex,\nstrongly convex, constrained, and unconstrained optimization problems. In each\ncase, we establish robust semi-global practical asymptotic or exponential\nstability results, and we show how to obtain well-posed discretized algorithms\nthat retain the main properties of the original dynamics. Given that existing\naveraging theorems for singularly perturbed hybrid systems are not directly\napplicable to our setting, we derive a new averaging theorem that relaxes some\nof the assumptions made in the literature, allowing us to make a clear link\nbetween the KL bounds that characterize the rates of convergence of the hybrid\ndynamics and their average dynamics. We also show that our results are\napplicable to non-hybrid algorithms, thus providing a general framework for\naccelerated dynamics based on averaging theory. We present different numerical\nexamples to illustrate our results.\n", "doi": null, "arxiv_id": "1909.00265", "pdf_urls": ["https://arxiv.org/pdf/1909.00265"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2020", "categories": "math.OC"} |
| {"title": "Model Reference Adaptive Control Allocation for Constrained Systems with\n Guaranteed Closed Loop Stability", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Seyed Shahabaldin Tohidi", "Yildiray Yildiz", "Ilya Kolmanovsky"], "abstract": " This paper proposes an adaptive control allocation approach for uncertain\nover-actuated systems with actuator saturation. The proposed method does not\nrequire uncertainty estimation or a persistent excitation assumption. Using the\nelement-wise non-symmetric projection algorithm, the adaptive parameters are\nrestricted to satisfy certain optimality conditions leading to overall closed\nloop system stability. Furthermore, a sliding mode controller with a\ntime-varying sliding surface, working in tandem with the adaptive control\nallocation, is proposed to guarantee the outer loop stability and reference\ntracking in the presence of control allocation errors and disturbances.\nSimulation results are provided, where the Aerodata Model in Research\nEnvironment is used as an over-actuated system with actuator saturation, to\ndemonstrate the effectiveness of the proposed method.\n", "doi": "10.1016/j.automatica.2020.109161", "arxiv_id": "1909.10036", "pdf_urls": ["https://arxiv.org/pdf/1909.10036"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 121 (2020)", "categories": "eess.SY cs.SY math.OC"} |
| {"title": "Robust Global Exponential Stabilization on the $n$-Dimensional Sphere\n with Applications to Trajectory Tracking for Quadrotors", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Pedro Casau and Christopher G. Mayhew and Ricardo G. Sanfelice and\n Carlos Silvestre"], "abstract": " In this paper, we design a hybrid controller that globally exponentially\nstabilizes a system evolving on the n-dimensional sphere, denoted by Sn. This\nhybrid controller is induced by a synergistic collection of potential functions\non Sn. We propose a particular construction of this class of functions that\ngenerates flows along geodesics of the sphere, providing convergence to the\ndesired reference with minimal path length. We show that the proposed strategy\nis suitable to the exponential stabilization of a quadrotor vehicle.\n", "doi": "10.1016/j.automatica.2019.108534", "arxiv_id": "1910.02026", "pdf_urls": ["https://arxiv.org/pdf/1910.02026"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 110 (2019)", "categories": "eess.SY cs.SY"} |
| {"title": "Pose Localization of Leader-Follower Networks with Direction\n Measurements", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Quoc Van Tran", "Brian D. O. Anderson", "and Hyo-Sung Ahn"], "abstract": " A distributed pose localization framework based on direction measurements is\nproposed for a type of \\textit{leader-follower} multi-agent systems in\n$\\mathbb{R}^3$. The novelty of the proposed localization method lies in the\nelimination of the need for using distance measurements and relative\norientation measurements for the network pose localization problem. In\nparticular, a network localization scheme is developed based directly on the\nmeasured direction constraints between an agent and its neighboring agents in\nthe network. The proposed position and orientation localization algorithms are\nimplemented through differential equations which simultaneously compute poses\nof all followers by using locally measured directional vectors and angular\nvelocities, and actual pose knowledge of some leader agents, allowing some\ntracking of time-varying orientations. Further, we establish an almost global\nasymptotic convergence of the estimated positions and orientations of the\nagents to the actual poses in the stationary case.\n", "doi": "10.1016/j.automatica.2020.109125.", "arxiv_id": "1910.03978", "pdf_urls": ["https://arxiv.org/pdf/1910.03978"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 120, 2020, article no. 109125", "categories": "math.OC"} |
| {"title": "A nonlinear tracking model predictive control scheme for dynamic target\n signals", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Johannes K\\\"ohler", "Matthias A. M\\\"uller", "Frank Allg\\\"ower"], "abstract": " We present a nonlinear model predictive control (MPC) scheme for tracking of\ndynamic target signals. The scheme combines stabilization and dynamic\ntrajectory planning in one layer, thus ensuring constraint satisfaction\nirrespective of changes in the dynamic target signal. For periodic target\nsignals we ensure exponential stability of the optimal reachable periodic\ntrajectory using suitable terminal ingredients and a convexity condition for\nthe underlying periodic optimal control problem. Furthermore, we introduce an\nonline optimization of the terminal set size to automate the trade-off between\nfast convergence and operation close to the constraints. In addition, we show\nhow stabilization and dynamic trajectory planning can be formulated as\npartially decoupled optimization problems, which reduces the computational\ndemand while ensuring recursive feasibility and convergence. The main tool to\nenable the proposed design is a novel reference generic offline computation\nthat provides suitable terminal ingredients for tracking of dynamic reference\ntrajectories. The practicality of this approach is demonstrated on benchmark\nexamples, which demonstrates superior performance compared to state of the art\napproaches.\n", "doi": "10.1016/j.automatica.2020.109030", "arxiv_id": "1911.03304", "pdf_urls": ["https://arxiv.org/pdf/1911.03304"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 118 (2020) 109030", "categories": "eess.SY cs.SY"} |
| {"title": "Structural Controllability of Networked Relative Coupling Systems", "year": 2019, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Yuan Zhang", "Yuanqing Xia", "Dihua Zhai"], "abstract": " This paper studies the controllability of networked relative coupling systems\n(NRCSs), in which subsystems are of fixed high-order linear dynamics and\ncoupled through relative variables depending on their neighbors, from a\nstructural perspective. The purpose is to explore conditions for subsystem\ndynamics and network topologies under which for almost all weights of the\nsubsystem interaction links, the corresponding numerical NRCSs are\ncontrollable, which is called structurally controllable. Three types of\nsubsystem interaction fashions are considered: 1) each subsystem is\nsingle-input-single-output (SISO), 2) each subsystem is\nmultiple-input-multiple-output (MIMO), and the weights for all channels between\ntwo subsystems are identical, and 3) each subsystem is MIMO, but different\nchannels between two subsystems can be weighted differently. {We show that all\nparameter-dependent modes of the NRCSs are generically controllable under some\nnecessary connectivity conditions. We then derive necessary and/or sufficient\nconditions for structural controllability depending on subsystem dynamics and\nnetwork topologies' connectivity in a decoupled form for all the three\ninteraction fashions.} We also extend our results to handle certain subsystem\nheterogeneities and demonstrate their direct applications on some practical\nsystems, including the mass-spring-damper system and the power network.\ndoiarxiv_id1911.06450pdf_urlshttps://arxiv.org/pdf/1911.06450githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2021categorieseess.SY cs.SY |
| titleBoundary feedback stabilization of a reaction-diffusion equation with\n Robin boundary conditions and state-delayyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsHugo Lhachemi and Robert Shortenabstract This paper discusses the boundary feedback stabilization of a\nreaction-diffusion equation with Robin boundary conditions and in the presence\nof a time-varying state-delay. The proposed control design strategy is based on\na finite-dimensional truncated model obtained via a spectral decomposition. By\nan adequate selection of the number of modes of the original\ninfinite-dimensional system, we show that the design performed on the\nfinite-dimensional truncated model achieves the exponential stabilization of\nthe original infinite-dimensional system. In the presence of distributed\ndisturbances, we show that the closed-loop system is exponentially\ninput-to-state stable with fading memory.\ndoi10.1016/j.automatica.2020.108931arxiv_id1911.10761pdf_urlshttps://arxiv.org/pdf/1911.10761githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 116, June 2020, 108931categoriesmath.OC cs.SY eess.SY |
| titleContinuous-time fully distributed generalized Nash equilibrium seeking\n for multi-integrator agentsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMattia Bianchi and Sergio Grammaticoabstract We consider strongly monotone games with convex separable coupling\nconstraints, played by dynamical agents, in a partial-decision information\nscenario. We start by designing continuous-time fully distributed feedback\ncontrollers, based on consensus and primal-dual gradient dynamics, to seek a\ngeneralized Nash equilibrium in networks of single-integrator agents. Our first\nsolution adopts a fixed gain, whose choice requires the knowledge of some\nglobal parameters of the game. To relax this requirement, we conceive a\ncontroller that can be tuned in a completely decentralized fashion, thanks to\nthe use of uncoordinated integral adaptive weights. We further introduce\nalgorithms specifically devised for generalized aggregative games. Finally, we\nadapt all our control schemes to deal with heterogeneous multi-integrator\nagents and, in turn, with nonlinear feedback-linearizable dynamical systems.\nFor all the proposed dynamics, we show convergence to a variational\nequilibrium, by leveraging monotonicity properties and stability theory for\nprojected dynamical systems.\ndoi10.1016/j.automatica.2021.109660arxiv_id1911.12266pdf_urlshttps://arxiv.org/pdf/1911.12266githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 129, July 2021categoriesmath.OC cs.GT cs.MA |
| titleDecomposed Structured Subsets for Semidefinite and Sum-of-Squares\n Optimizationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsJared MillerYang ZhengMario SznaierAntonis Papachristodoulouabstract Semidefinite programs (SDPs) are standard convex problems that are frequently\nfound in control and optimization applications. Interior-point methods can\nsolve SDPs in polynomial time up to arbitrary accuracy, but scale poorly as the\nsize of matrix variables and the number of constraints increases. To improve\nscalability, SDPs can be approximated with lower and upper bounds through the\nuse of structured subsets (e.g., diagonally-dominant and scaled-diagonally\ndominant matrices). Meanwhile, any underlying sparsity or symmetry structure\nmay be leveraged to form an equivalent SDP with smaller positive semidefinite\nconstraints. In this paper, we present a notion of decomposed structured\nsubsets}to approximate an SDP with structured subsets after an equivalent\nconversion. The lower/upper bounds found by approximation after conversion\nbecome tighter than the bounds obtained by approximating the original SDP\ndirectly. We apply decomposed structured subsets to semidefinite and\nsum-of-squares optimization problems with examples of H-infinity norm\nestimation and constrained polynomial optimization. An existing basis pursuit\nmethod is adapted into this framework to iteratively refine bounds.\ndoi10.1016/j.automatica.2021.110125arxiv_id1911.12859pdf_urlshttps://arxiv.org/pdf/1911.12859githubhttps://github.com/zhengy09/SDPfwhttps://github.com/soc-ucsd/sdpfwis_oasourcesopen-arxivtypejournaljournal_refAutomatica 137 (2022) 110125categoriesmath.OC cs.SY eess.SY |
| titleNetwork games with dynamic players: Stabilization and output convergence\n to Nash equilibriumyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMeichen Guo and Claudio De Persisabstract This paper addresses a class of network games played by dynamic agents using\ntheir outputs. Unlike most existing related works, the Nash equilibrium in this\nwork is defined by functions of agent outputs instead of full agent states,\nwhich allows the agents to have more general and heterogeneous dynamics and\nmaintain some privacy of their local states. The concerned network game is\nformulated with agents modeled by uncertain linear systems subject to external\ndisturbances. The cost function of each agent is a linear quadratic function\ndepending on the outputs of its own and its neighbors in the underlying graph.\nThe main challenge stemming from this game formulation is that merely driving\nthe agent outputs to the Nash equilibrium does not guarantee the stability of\nthe agent dynamics. Using local output and the outputs from the neighbors of\neach agent, we aim at designing game strategies that achieve output Nash\nequilibrium seeking and stabilization of the closed-loop dynamics.\nParticularly, when each agents knows how the actions of its neighbors affect\nits cost function, a game strategy is developed for network games with digraph\ntopology. When each agent is also allowed to exchange part of its compensator\nstate, a distributed strategy can be designed for networks with connected\nundirected graphs or connected digraphs.\ndoi10.1016/j.automatica.2021.109711arxiv_id1912.00383pdf_urlshttps://arxiv.org/pdf/1912.00383githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 130 (2021) 109711categoriesmath.OC cs.SY eess.SY |
| titleEquilibrium strategies in a multiregional transboundary pollution\n differential game with spatially distributed controlsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsJavier de FrutosPaula L\\'opez and Guiomar Mart\\'in-Herr\\'an"], "abstract": " We analyze a differential game with spatially distributed controls to study a\nmultiregional transboundary pollution problem. The dynamics of the state\nvariable (pollution stock) is defined by a two dimensional parabolic partial\ndifferential equation. The control variables (emissions) are spatially\ndistributed variables. The model allows for a, possibly large, number of agents\nwith predetermined geographical relationships. For a special functional form\npreviously used in the literature of transboundary pollution dynamic games we\nanalytically characterize the feedback Nash equilibrium. We show that at the\nequilibrium both the level and the location of emissions of each region depend\non the particular geographical relationship among agents. We prove that, even\nin a simplified model, the geographical considerations can modify the players'\noptimal strategies and therefore, the spatial aspects of the model should not\nbe overlooked.\ndoi10.1016/j.automatica.2020.109411arxiv_id1912.00481pdf_urlshttps://arxiv.org/pdf/1912.00481githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, March 2021, 109411categoriesmath.OC |
| titleLocal Stability of PD Controlled Bipedal Walking Robotsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsShishir Kolathayaabstract We establish stability results for PD tracking control laws in bipedal\nwalking robots. Stability of PD control laws for continuous robotic systems is\nan established result, and we extend this for hybrid robotic systems, an\nalternating sequence of continuous and discrete events. Bipedal robots have the\nleg-swing as the continuous event, and the foot-strike as the discrete event.\nIn addition, bipeds largely have underactuations due to the interactions\nbetween feet and ground. For each continuous event, we establish that the\nconvergence rate of the tracking error can be regulated via appropriate tuning\nof the PD gains; and for each discrete event, we establish that this\nconvergence rate sufficiently overcomes the nonlinear impacts by assumptions on\nthe hybrid zero dynamics. The main contributions are 1) Extension of the\nstability results of PD control laws for underactuated robotic systems, and 2)\nExponential ultimate boundedness of hybrid periodic orbits under the assumption\nof exponential stability of their projections to the hybrid zero dynamics.\nTowards the end, we will validate these results in a 2-link bipedal walker in\nsimulation.\ndoi10.1016/j.automatica.2020.108841arxiv_id2001.00145pdf_urlshttps://arxiv.org/pdf/2001.00145githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 114 (2020) 108841categoriescs.RO cs.SY eess.SY |
| titleStability of Multi-Dimensional Switched Systems with an Application to\n Open Multi-Agent Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMengqi XueYang TangWei Renand Feng Qianabstract A multi-dimensional switched system or multi-mode multi-dimensional ($M^3D$)\nsystem extends the classic switched system by allowing different subsystem\ndimensions. The stability problem of the $M^3D$ system, whose state transitions\nat switching instants can be discontinuous due to the dimension-varying\nfeature, is studied. The discontinuous state transition is formulated by an\naffine map that captures both the dimension variations and the state impulses,\nwith no extra constraint imposed. In the presence of unstable subsystems, the\ngeneral criteria featuring a series of Lyapunov-like conditions for the\npractical and asymptotic stability properties of the $M^3D$ system are provided\nunder the proposed slow/fast transition-dependent average dwell time framework.\nThen, by considering linear subsystems, we propose a class of parametric\nmultiple Lyapunov functions to verify the obtained Lyapunov-like stability\nconditions and explicitly reveal a connection between the practical/asymptotic\nstability property and the non-vanishing/vanishing property of the impulsive\neffects in the state transition process. Further, the obtained stability\nresults for the $M^3D$ system are applied to the consensus problem of the open\nmulti-agent system (MAS), whose network topology can be switching and\nsize-varying due to the migrations of agents. It shows that through a proper\ntransformation, the seeking of the (practical) consensus performance of the\nopen MAS with disconnected digraphs boils down to that of the (practical)\nstability property of an $M^3D$ system with unstable subsystems.\ndoi10.1016/j.automatica.2022.110644arxiv_id2001.00435pdf_urlshttps://arxiv.org/pdf/2001.00435githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 146 (2022) 110644categorieseess.SY cs.SY |
| titleOn the confidentiality of controller states under sensor attacksyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsDavid UmsonstHenrik Sandbergabstract With the emergence of cyber-attacks on control systems it has become clear\nthat improving the security of control systems is an important task in today's\nsociety. We investigate how an attacker that has access to the measurements\ntransmitted from the plant to the controller can perfectly estimate the\ninternal state of the controller. This attack on sensitive information of the\ncontrol loop is, on the one hand, a violation of the privacy, and, on the other\nhand, a violation of the security of the closed-loop system if the obtained\nestimate is used in a larger attack scheme. Current literature on sensor\nattacks often assumes that the attacker has already access to the controller's\nstate. However, this is not always possible. We derive conditions for when the\nattacker is able to perfectly estimate the controller's state. These conditions\nshow that if the controller has unstable poles a perfect estimate of the\ncontroller state is not possible. Moreover, we propose a defence mechanism to\nrender the attack infeasible. This defence is based on adding uncertainty to\nthe controller dynamics. We also discuss why an unstable controller is only a\ngood defence for certain plants. Finally, simulations with a three-tank system\nverify our results.\n", "doi": "10.1016/j.automatica.2020.109329", "arxiv_id": "2001.02714", "pdf_urls": ["https://arxiv.org/pdf/2001.02714"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 123,2021", "categories": "eess.SY cs.SY math.OC"} |
| {"title": "Robustness of constant-delay predictor feedback for in-domain\n stabilization of reaction-diffusion PDEs with time- and spatially-varying\n input delays", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Hugo Lhachemi", "Christophe Prieur and Robert Shorten"], "abstract": " This paper discusses the in-domain feedback stabilization of\nreaction-diffusion PDEs with Robin boundary conditions in the presence of an\nuncertain time- and spatially-varying delay in the distributed actuation. The\nproposed control design strategy consists of a constant-delay predictor\nfeedback designed based on the known nominal value of the control input delay\nand is synthesized on a finite-dimensional truncated model capturing the\nunstable modes of the original infinite-dimensional system. By using a\nsmall-gain argument, we show that the resulting closed-loop system is\nexponentially stable provided that the variations of the delay around its\nnominal value are small enough. The proposed proof actually applies to any\ndistributed-parameter system associated with an unbounded operator that 1)\ngenerates a $C_0$-semigroup on a weighted space of square integrable functions\nover a compact interval; and 2) is self-adjoint with compact resolvent.\n", "doi": "10.1016/j.automatica.2020.109347", "arxiv_id": "2002.09200", "pdf_urls": ["https://arxiv.org/pdf/2002.09200"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2021, vol. 123, p. 109347", "categories": "math.OC cs.SY eess.SY"} |
| {"title": "Feedback Identification of conductance-based models", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Thiago B. Burghi", "Maarten Schoukens", "Rodolphe Sepulchre"], "abstract": " This paper applies the classical prediction error method (PEM) to the\nestimation of nonlinear discrete-time models of neuronal systems subject to\ninput-additive noise. While the nonlinear system exhibits excitability,\nbifurcations, and limit-cycle oscillations, we prove consistency of the\nparameter estimation procedure under output feedback. Hence, this paper\nprovides a rigorous framework for the application of conventional nonlinear\nsystem identification methods to discrete-time stochastic neuronal systems. The\nmain result exploits the elementary property that conductance-based models of\nneurons have an exponentially contracting inverse dynamics. This property is\nimplied by the voltage-clamp experiment, which has been the fundamental\nmodeling experiment of neurons ever since the pioneering work of Hodgkin and\nHuxley.\n", "doi": "10.1016/j.automatica.2020.109297", "arxiv_id": "2002.09626", "pdf_urls": ["https://arxiv.org/pdf/2002.09626"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica; 2021. Vol 123. Article 109297", "categories": "eess.SY cs.SY q-bio.NC"} |
| {"title": "Uncertainty estimation in equality-constrained MAP and maximum\n likelihood estimation with applications to system identification and state\n estimation", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Dimas Abreu Archanjo Dutra"], "abstract": " In unconstrained maximum a posteriori (MAP) and maximum likelihood\nestimation, the inverse of minus the merit-function Hessian matrix is an\napproximation of the estimate covariance matrix. In the Bayesian context of MAP\nestimation, it is the covariance of a normal approximation of the posterior\naround the mode; while in maximum likelihood estimation, it an approximation of\nthe inverse Fisher information matrix, to which the covariance of efficient\nestimators converge. These measures are routinely used in system identification\nto evaluate the estimate uncertainties and diagnose problems such as\noverparametrization, improper excitation and unidentifiability. A wide variety\nof estimation problems in systems and control, however, can be formulated as\nequality-constrained optimizations with additional decision variables to\nexploit parallelism in computer hardware, simplify implementation and increase\nthe convergence basin and efficiency of the nonlinear program solver. The\nintroduction of the extra variables, however, dissociates the inverse Hessian\nfrom the covariance matrix. Instead, submatrices of the inverse Hessian of the\nconstrained-problem's Lagrangian must be used. In this paper, we derive these\nrelationships, showing how the estimates' covariance can be estimated directly\nfrom the augmented problem. Application examples are shown in system\nidentification with the output-error method and joint state-path and parameter\nestimation.\n", "doi": "10.1016/j.automatica.2020.108935", "arxiv_id": "2002.10975", "pdf_urls": ["https://arxiv.org/pdf/2002.10975"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 116 (June 2020), 108935", "categories": "stat.ME math.OC"} |
| {"title": "Periodic event-triggered output regulation for linear multi-agent\n systems", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Shiqi Zheng", "Peng Shi", "Ramesh K. Agarwal", "Chee Peng Lim"], "abstract": " This study considers the problem of periodic event-triggered (PET)\ncooperative output regulation for a class of linear multi-agent systems. The\nadvantage of the PET output regulation is that the data transmission and\ntriggered condition are only needed to be monitored at discrete sampling\ninstants. It is assumed that only a small number of agents can have access to\nthe system matrix and states of the leader. Meanwhile, the PET mechanism is\nconsidered not only in the communication between various agents, but also in\nthe sensor-to-controller and controller-to-actuator transmission channels for\neach agent. The above problem set-up will bring some challenges to the\ncontroller design and stability analysis. Based on a novel PET distributed\nobserver, a PET dynamic output feedback control method is developed for each\nfollower. Compared with the existing works, our method can naturally exclude\nthe Zeno behavior, and the inter-event time becomes multiples of the sampling\nperiod. Furthermore, for every follower, the minimum inter-event time can be\ndetermined \\textit{a prior}, and computed directly without the knowledge of the\nleader information. An example is given to verify and illustrate the\neffectiveness of the new design scheme.\n", "doi": "10.1016/j.automatica.2020.109223", "arxiv_id": "2003.03560", "pdf_urls": ["https://arxiv.org/pdf/2003.03560"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2020", "categories": "eess.SY cs.SY"} |
| {"title": "Optimal Composition of Heterogeneous Multi-Agent Teams for Coverage\n Problems with Performance Bound Guarantees", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Chuangchuang Sun and Shirantha Welikala and Christos G. Cassandras"], "abstract": " We consider the problem of determining the optimal composition of a\nheterogeneous multi-agent team for coverage problems by including costs\nassociated with different agents and subject to an upper bound on the maximal\nallowable number of agents. We formulate a resource allocation problem without\nintroducing additional non-convexities to the original problem. We develop a\ndistributed Projected Gradient Ascent (PGA) algorithm to solve the optimal team\ncomposition problem. To deal with non-convexity, we initialize the algorithm\nusing a greedy method and exploit the submodularity and curvature properties of\nthe coverage objective function to derive novel tighter performance bound\nguarantees on the optimization problem solution. Numerical examples are\nincluded to validate the effectiveness of this approach in diverse mission\nspace configurations and different heterogeneous multi-agent collections.\nComparative results obtained using a commercial mixed-integer nonlinear\nprogramming problem solver demonstrate both the accuracy and computational\nefficiency of the distributed PGA algorithm.\n", "doi": "10.1016/j.automatica.2020.108961", "arxiv_id": "2003.05587", "pdf_urls": ["https://arxiv.org/pdf/2003.05587"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica (2020) 117", "categories": "eess.SY cs.SY math.OC"} |
| {"title": "Cohesive Networks using Delayed Self Reinforcement", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Santosh Devasia"], "abstract": " How a network gets to the goal (a consensus value) can be as important as\nreaching the consensus value. While prior methods focus on rapidly getting to a\nnew consensus value, maintaining cohesion, during the transition between\nconsensus values or during tracking, remains challenging and has not been\naddressed. The main contributions of this work are to address the problem of\nmaintaining cohesion by: (i) proposing a new delayed self-reinforcement (DSR)\napproach; (ii) extending it for use with agents that have higher-order,\nheterogeneous dynamics, and (iii) developing stability conditions for the\nDSR-based method. With DSR, each agent uses current and past information from\nneighbors to infer the overall goal and modifies the update law to improve\ncohesion. The advantages of the proposed DSR approach are that it only requires\nalready-available information from a given network to improve the cohesion and\ndoes not require network-connectivity modifications (which might not be always\nfeasible) nor increases in the system's overall response speed (which can\nrequire larger input). Moreover, illustrative simulation examples are used to\ncomparatively evaluate the performance with and without DSR. The simulation\nresults show substantial improvement in cohesion with DSR.\ndoi10.1016/j.automatica.2019.108699arxiv_id2003.06679pdf_urlshttps://arxiv.org/pdf/2003.06679githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 112, 2020, 108699categorieseess.SY cs.MA cs.SY |
| titleIndefinite Mean-Field Type Linear-Quadratic Stochastic Optimal Control\n Problemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsNa LiXun Li and Zhiyong Yuabstract This paper focuses on indefinite stochastic mean-field linear-quadratic\n(MF-LQ, for short) optimal control problems, which allow the weighting matrices\nfor state and control in the cost functional to be indefinite. The solvability\nof stochastic Hamiltonian system and Riccati equations is presented under both\npositive definite case and indefinite case. The optimal controls in open-loop\nform and closed-loop form are obtained, respectively. Moreover, the dynamic\nmean-variance problem can be solved within the framework of the indefinite\nMF-LQ problem. Other two examples shed light on the theoretical results\nestablished.\ndoi10.1016/j.automatica.2020.109267arxiv_id2003.08090pdf_urlshttps://arxiv.org/pdf/2003.08090githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 122 (2020) 109267categoriesmath.OC |
| titleA Predictor-Corrector Type Algorithm for the Pseudospectral Abscissa\n Computation of Time-Delay Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsSuat GumussoyWim Michielsabstract The pseudospectrum of a linear time-invariant system is the set in the\ncomplex plane consisting of all the roots of the characteristic equation when\nthe system matrices are subjected to all possible perturbations with a given\nupper bound. The pseudospectral abscissa is defined as the maximum real part of\nthe characteristic roots in the pseudospectrum and, therefore, it is for\ninstance important from a robust stability point of view. In this paper we\npresent an accurate method for the computation of the pseudospectral abscissa\nof retarded delay differential equations with discrete pointwise delays. Our\napproach is based on the connections between the pseudospectrum and the level\nsets of an appropriately defined complex function. The computation is done in\ntwo steps. In the prediction step, an approximation of the pseudospectral\nabscissa is obtained based on a rational approximation of the characteristic\nmatrix and the application of a bisection algorithm. Each step in this\nbisection algorithm relies on checking the presence of the imaginary axis\neigenvalues of a complex matrix, similar to the delay free case. In the\ncorrector step, the approximate pseudospectral abscissa is corrected to any\ngiven accuracy, by solving a set of nonlinear equations that characterize\nextreme points in the pseudospectrum contours.\ndoi10.1016/j.automatica.2010.01.032arxiv_id2003.08297pdf_urlshttps://arxiv.org/pdf/2003.08297githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, volume 46, issue 4, 2010, pages 657-664categorieseess.SY cs.SY |
| titleOn the capability of a class of quantum sensorsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsQi YuYuanlong WangDaoyi DongIan R. Petersenabstract Quantum sensors may provide extremely high sensitivity and precision to\nextract key information in a quantum or classical physical system. A\nfundamental question is whether a quantum sensor is capable of uniquely\ninferring unknown parameters in a system for a given structure of the quantum\nsensor and admissible measurement on the sensor. In this paper, we investigate\nthe capability of a class of quantum sensors which consist of either a single\nqubit or two qubits. A quantum sensor is coupled to a spin chain system to\nextract information of unknown parameters in the system. With given\ninitialisation and measurement schemes, we employ the similarity transformation\napproach and the Grobner basis method to prove that a single-qubit quantum\nsensor cannot effectively estimate the unknown parameters in the spin chain\nsystem while the two-qubit quantum sensor can. The work demonstrates that it is\na feasible method to enhance the capability of quantum sensors by increasing\nthe number of qubits in the quantum sensors for some practical applications.\ndoi10.1016/j.automatica.2021.109612arxiv_id2003.08679pdf_urlshttps://arxiv.org/pdf/2003.08679githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2021, vol. 129, p.109612categoriesquant-ph |
| titleRoot Locus for SISO Dead-Time Systems: A Continuation Based Approachyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsSuat GumussoyWim Michielsabstract We present a numerical method to plot the root locus of\nSingle-Input-Single-Output (SISO) dead-time systems with respect to the\ncontroller gain or the system delay. We compute the trajectories of\ncharacteristic roots of the closed-loop system on a prescribed complex right\nhalf-plane. We calculate the starting, branch and boundary crossing roots of\nroot-locus branches inside the region. We compute the root locus of each\ncharacteristic root based on a predictor-corrector type continuation method. To\navoid the high sensitivity of roots with respect to the locus parameter in the\nneighborhood of branch points, the continuation method relies on a natural\nparameterization of the root-locus trajectory in terms of a distance in the\n(characteristic root, locus parameter)-space. The method is numerically stable\nfor high order SISO dead-time systems.\ndoi10.1016/j.automatica.2012.01.010arxiv_id2003.09253pdf_urlshttps://arxiv.org/pdf/2003.09253githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, volume 48, issue 3, 2012categorieseess.SY cs.SY |
| titleFast generalized Nash equilibrium seeking under partial-decision\n informationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsMattia BianchiGiuseppe BelgioiosoSergio Grammaticoabstract We address the generalized Nash equilibrium seeking problem in a\npartial-decision information scenario, where each agent can only exchange\ninformation with some neighbors, although its cost function possibly depends on\nthe strategies of all agents. The few existing methods build on projected\npseudo-gradient dynamics, and require either double-layer iterations or\nconservative conditions on the step sizes. To overcome both these flaws and\nimprove efficiency, we design the first fully-distributed single-layer\nalgorithms based on proximal best-response. Our schemes are fixed-step and\nallow for inexact updates, which is crucial for reducing the computational\ncomplexity. Under standard assumptions on the game primitives, we establish\nconvergence to a variational equilibrium (with linear rate for games without\ncoupling constraints) by recasting our algorithms as proximal-point methods,\nopportunely preconditioned to distribute the computation among the agents.\nSince our analysis hinges on a restricted monotonicity property, we also\nprovide new general results that significantly extend the domain of\napplicability of proximal-point methods. Besides, the operator-theoretic\napproach favors the implementation of provably correct acceleration schemes\nthat can further improve the convergence speed. Finally, the potential of our\nalgorithms is demonstrated numerically, revealing much faster convergence with\nrespect to projected pseudo-gradient methods and validating our theoretical\nfindings.\ndoi10.1016/j.automatica.2021.110080arxiv_id2003.09335pdf_urlshttps://arxiv.org/pdf/2003.09335githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 136, 2022, 110080, ISSN 0005-1098,categoriesmath.OC cs.GT cs.MA |
| titleComputation of Extremum Singular Values and the Strong H-infinity Norm\n of SISO Time-Delay Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsSuat GumussoyWim Michielsabstract We consider the computation of H-infinity norms for\nSingle-Input-Single-Output (SISO) time-delay systems, which are described by\ndelay differential algebraic equations. Unlike the iterative level set methods\nin the literature, we present a novel numerical method to compute the\nH-infinity norm. This method requires solving one eigenvalue problem of at most\ntwice the size of the eigenvalue problem in every iteration of a level set\nmethod, but in practice often considerably lower. We first show that the\ncomputation of extrema of the transfer function can be turned into the\ncomputation of the imaginary axis zeros of a transcendental function. We\ncompute these zeros by a predictor-corrector type algorithm. It is known that\nthe H-infinity norm of delay differential algebraic systems, which can model\nboth retarded and neutral type systems, might be sensitive with respect to\narbitrarily small delay perturbations. This recently led to the concept of\nstrong H-infinity norms, which explicitly take into account such small delay\nperturbations. We present a direct numerical method to compute the strong\nH-infinity norm of SISO time-delay systems. Our algorithm is applicable to the\nclosed-loop system of interconnections (series, parallel, feedback, junctions)\nof time-delay systems and/or controllers.\ndoi10.1016/j.automatica.2015.02.003arxiv_id2003.10048pdf_urlshttps://arxiv.org/pdf/2003.10048githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, volume 54, pages 266-271, 2015categorieseess.SY cs.SY |
| titleOptimal Time Trajectory and Coordination for Connected and Automated\n Vehiclesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAndreas A. MalikopoulosLogan E. BeaverIoannis Vasileios Chremosabstract In this paper, we provide a decentralized theoretical framework for\ncoordination of connected and automated vehicles (CAVs) at different traffic\nscenarios. The framework includes: (1) an upper-level optimization that yields\nfor each CAV its optimal time trajectory and lane to pass through a given\ntraffic scenario while alleviating congestion; and (2) a low-level optimization\nthat yields for each CAV its optimal control input (acceleration/deceleration).\nWe provide a complete, analytical solution of the low-level optimization\nproblem that includes the rear-end, speed-dependent safety constraint.\nFurthermore, we provide a problem formulation for the upper-level optimization\nin which there is no duality gap. The latter implies that the optimal time\ntrajectory for each CAV does not activate any of the state, control, and safety\nconstraints of the low-level optimization, thus allowing for online\nimplementation. Finally, we present a geometric duality framework with\nhyperplanes to derive the condition under which the optimal solution of the\nupper-level optimization always exists. We validate the effectiveness of the\nproposed theoretical framework through simulation.\ndoi10.1016/j.automatica.2020.109469arxiv_id2003.12183pdf_urlshttps://arxiv.org/pdf/2003.12183githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 125, 2021categoriesmath.OC |
| titleCompositional Synthesis of Opacity-Preserving Finite Abstractions for\n Interconnected Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsSiyuan Liu and Majid Zamaniabstract In this paper, we propose a compositional approach to construct\nopacity-preserving finite abstractions (a.k.a symbolic models) for networks of\ndiscrete-time nonlinear control systems. Particularly, we introduce new notions\nof simulation functions that characterize the distance between control systems\nwhile preserving opacity properties across them. Instead of treating\nlarge-scale systems in a monolithic manner, we develop a compositional scheme\nto construct the interconnected finite abstractions together with the overall\nopacity-preserving simulation functions. For a network of incrementally\ninput-to-state stable control systems and under some small-gain type condition,\nan algorithm for designing local quantization parameters is presented to\norderly build the local symbolic models of subsystems such that the network of\nsymbolic models simulates the original network for an a-priori defined accuracy\nwhile preserving its opacity properties.\ndoi10.1016/j.automatica.2021.109745arxiv_id2004.00131pdf_urlshttps://arxiv.org/pdf/2004.00131githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 131, September 2021categorieseess.SY cs.SY |
| titleResidual-driven Fuzzy C-Means Clustering for Image Segmentationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsCong WangWitold PedryczZhiWu LiMengChu Zhouabstract Due to its inferior characteristics, an observed (noisy) image's direct use\ngives rise to poor segmentation results. Intuitively, using its noise-free\nimage can favorably impact image segmentation. Hence, the accurate estimation\nof the residual between observed and noise-free images is an important task. To\ndo so, we elaborate on residual-driven Fuzzy C-Means (FCM) for image\nsegmentation, which is the first approach that realizes accurate residual\nestimation and leads noise-free image to participate in clustering. We propose\na residual-driven FCM framework by integrating into FCM a residual-related\nfidelity term derived from the distribution of different types of noise. Built\non this framework, we present a weighted $\\ell_{2}$-norm fidelity term by\nweighting mixed noise distribution, thus resulting in a universal\nresidual-driven FCM algorithm in presence of mixed or unknown noise. Besides,\nwith the constraint of spatial information, the residual estimation becomes\nmore reliable than that only considering an observed image itself. Supporting\nexperiments on synthetic, medical, and real-world images are conducted. The\nresults demonstrate the superior effectiveness and efficiency of the proposed\nalgorithm over existing FCM-related algorithms.\n", "doi": "10.1109/jas.2020.1003420", "arxiv_id": "2004.07160", "pdf_urls": ["https://arxiv.org/pdf/2004.07160"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "IEEE/CAA Journal of Automatica Sinica, 2020", "categories": "eess.IV cs.CV"} |
| {"title": "A Lyapunov Function for the Combined System-Optimizer Dynamics in\n Inexact Model Predictive Control", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Andrea Zanelli", "Quoc Tran Dinh", "Moritz Diehl"], "abstract": " In this paper, an asymptotic stability proof for a class of methods for\ninexact nonlinear model predictive control is presented. General Q-linearly\nconvergent online optimization methods are considered and an asymptotic\nstability result is derived for the setting where a limited number of\niterations of the optimizer are carried out per sampling time. Under the\nassumption of Lipschitz continuity of the solution, we explicitly construct a\nLyapunov function for the combined system-optimizer dynamics, which shows that\nasymptotic stability can be obtained if the sampling time is sufficiently\nshort. The results constitute an extension to existing attractivity results\nwhich hold in the simplified setting where inequality constraints are either\nnot present or inactive in the region of attraction considered. Moreover, with\nrespect to the established results on robust asymptotic stability of suboptimal\nmodel predictive control, we develop a framework that takes into account the\noptimizer's dynamics and does not require decrease of the objective function\nacross iterates. By extending these results, the gap between theory and\npractice of the standard real-time iteration strategy is bridged and asymptotic\nstability for a broader class of methods is guaranteed.\ndoiarxiv_id2004.08578pdf_urlshttps://arxiv.org/pdf/2004.08578githubhttps://github.com/zanellia/nmpc_system_optimizer_lyapunovis_oasourcesopen-arxivtypejournaljournal_refAutomatica 2021categoriesmath.OC |
| titleGeneralization of Affine Feedback Stock Trading Results to Include\n Stop-Loss Ordersyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsChung-Han Hsiehabstract The takeoff point of this paper is to generalize the existing stock trading\nresults for a class of affine feedback controller to include consideration of a\nstop-loss order. Using the geometric Brownian motion as the underlying stock\nprice model, our main result is to provide a closed-form expression for the\ncumulative distribution function for the trading profit or loss. In addition,\nwe show that the affine feedback controller with stop-loss order indeed\ngeneralizes the result without stop order in the sense of distribution\nfunction. Some simulations and illustrative examples are also provided as\nsupporting evidence of the theory. Moreover, we provide some technical results\naimed at addressing the issues about survivability, cash-financing\nconsiderations, long-only property, and lower bound of the expected gain or\nloss.\ndoi10.1016/j.automatica.2021.110051arxiv_id2004.12848pdf_urlshttps://arxiv.org/pdf/2004.12848githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 136, pp. 110051:1-110051:7, 2022categoriesmath.OC cs.SY eess.SY q-fin.MF q-fin.RM |
| titleReduced-Dimensional Reinforcement Learning Control using Singular\n Perturbation Approximationsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsSayak MukherjeeHe BaiAranya Chakraborttyabstract We present a set of model-free, reduced-dimensional reinforcement learning\n(RL) based optimal control designs for linear time-invariant singularly\nperturbed (SP) systems. We first present a state-feedback and output-feedback\nbased RL control design for a generic SP system with unknown state and input\nmatrices. We take advantage of the underlying time-scale separation property of\nthe plant to learn a linear quadratic regulator (LQR) for only its slow\ndynamics, thereby saving a significant amount of learning time compared to the\nconventional full-dimensional RL controller. We analyze the sub-optimality of\nthe design using SP approximation theorems and provide sufficient conditions\nfor closed-loop stability. Thereafter, we extend both designs to clustered\nmulti-agent consensus networks, where the SP property reflects through\nclustering. We develop both centralized and cluster-wise block-decentralized RL\ncontrollers for such networks, in reduced dimensions. We demonstrate the\ndetails of the implementation of these controllers using simulations of\nrelevant numerical examples and compare them with conventional RL designs to\nshow the computational benefits of our approach.\ndoi10.1016/j.automatica.2020.109451arxiv_id2004.14501pdf_urlshttps://arxiv.org/pdf/2004.14501githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 2021 (full version with proofs)categorieseess.SY cs.LG cs.SY |
| titleOn a probabilistic approach to synthesize control policies from example\n datasetsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsDavide GagliardiGiovanni Russoabstract This paper is concerned with the design of control policies from example\ndatasets. The case considered is when just a black box description of the\nsystem to be controlled is available and the system is affected by actuation\nconstraints. These constraints are not necessarily fulfilled by the (possibly,\nnoisy) example data and the system under control is not necessarily the same as\nthe one from which these data are collected. In this context, we introduce a\nnumber of theoretical results to compute a control policy from example datasets\nthat: (i) makes the behavior of the closed-loop system similar to the one\nillustrated in the data; (ii) guarantees compliance with the constraints. We\nrecast the control problem as a finite-horizon optimal control problem and give\nan explicit expression for its optimal solution. Moreover, we turn our findings\ninto an algorithmic procedure. The procedure gives a systematic tool to compute\nthe policy. The effectiveness of our approach is illustrated via a numerical\nexample, where we use real data collected from test drives to synthesize a\ncontrol policy for the merging of a car on a highway.\ndoi10.1016/j.automatica.2021.110121arxiv_id2005.11191pdf_urlshttps://arxiv.org/pdf/2005.11191githubhttps://github.com/unisa-acg/discrete-fpdis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Vol. 137, 2022categoriesmath.OC |
| titleRobust exact differentiators with predefined convergence timeyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsRichard SeeberHernan HaimovichMartin HornLeonid FridmanHern\\'an De Battista"], "abstract": " The problem of exactly differentiating a signal with bounded second\nderivative is considered. A class of differentiators is proposed, which\nconverge to the derivative of such a signal within a fixed, i.e., a finite and\nuniformly bounded convergence time. A tuning procedure is derived that allows\nto assign an arbitrary, predefined upper bound for this convergence time. It is\nfurthermore shown that this bound can be made arbitrarily tight by appropriate\ntuning. The usefulness of the procedure is demonstrated by applying it to the\nwell-known uniform robust exact differentiator, which is included in the\nconsidered class of differentiators as a special case.\n", "doi": "10.1016/j.automatica.2021.109858", "arxiv_id": "2005.12366", "pdf_urls": ["https://arxiv.org/pdf/2005.12366"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, volume 134 (2021), article 109858", "categories": "eess.SY cs.SY math.DS"} |
| {"title": "Constructive Observer Design for Visual Simultaneous Localisation and Mapping", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Pieter van Goor", "Robert Mahony", "Tarek Hamel", "Jochen Trumpf"], "abstract": "Visual Simultaneous Localisation and Mapping (VSLAM) is a well-known problem in robotics with a large range of applications. This paper presents a novel approach to VSLAM by lifting the observer design to a novel Lie group on which the system output is equivariant. The perspective gained from this analysis facilitates the design of a non-linear observer with almost semi-globally asymptotically stable error dynamics. Simulations are provided to illustrate the behaviour of the proposed observer and experiments on data gathered using a fixed-wing UAV flying outdoors demonstrate its performance.", "doi": "10.1016/j.automatica.2021.109803", "arxiv_id": "2006.05053", "pdf_urls": ["https://arxiv.org/pdf/2006.05053"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 132 (2021): 109803", "categories": "eess.SY cs.SY"} |
| {"title": "Max-Min Lyapunov Functions for Switched Systems and Related Differential\n Inclusions", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Matteo Della Rossa and Aneel Tanwani and Luca Zaccarian"], "abstract": " Starting from a finite family of continuously differentiable positive\ndefinite functions, we study conditions under which a function obtained by\nmax-min combinations is a Lyapunov function, establishing stability for two\nkinds of nonlinear dynamical systems: a) Differential inclusions where the\nset-valued right-hand-side comprises the convex hull of a finite number of\nvector fields, and b) Autonomous switched systems with a state-dependent\nswitching signal. We investigate generalized notions of directional derivatives\nfor these max-min functions, and use them in deriving stability conditions with\nvarious degrees of conservatism, where more conservative conditions are\nnumerically more tractable. The proposed constructions also provide nonconvex\nLyapunov functions, which are shown to be useful for systems with\nstate-dependent switching that do not admit a convex Lyapunov function. Several\nexamples are included to illustrate the results.\n", "doi": "10.1016/j.automatica.2020.109123", "arxiv_id": "2006.07227", "pdf_urls": ["https://arxiv.org/pdf/2006.07227"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, volume 120, October 2020", "categories": "math.OC"} |
| {"title": "Variation diminishing linear time-invariant systems", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Christian Grussler and Rodolphe Sepulchre"], "abstract": " This paper studies the variation diminishing property of $k$-positive linear\ntime-invariant (LTI) systems, which map inputs with $k-1$ sign changes to\noutputs with at most the same variation. We characterize this property for the\nToeplitz and Hankel operators of finite-dimensional systems. Our main result is\nthat these operators have a dominant approximation in the form of series or\nparallel interconnections of $k$ first order positive systems. This is shown by\nexpressing the $k$-positivity of a LTI system as the external positivity (that\nis, $1$-positivity) of $k$ compound LTI systems. Our characterization\ngeneralizes well known properties of externally positive systems ($k=1$) and\ntotally positive systems ($k=\\infty$; also known as relaxation systems).\n", "doi": "10.1016/j.automatica.2021.109985", "arxiv_id": "2006.10030", "pdf_urls": ["https://arxiv.org/pdf/2006.10030"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 136, pp. 109985, 2022", "categories": "math.OC math.ST stat.TH"} |
| {"title": "Structure-Preserving Interpolation for Model Reduction of Parametric\n Bilinear Systems", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Peter Benner", "Serkan Gugercin", "Steffen W. R. Werner"], "abstract": " In this paper, we present an interpolation framework for structure-preserving\nmodel order reduction of parametric bilinear dynamical systems. We introduce a\ngeneral setting, covering a broad variety of different structures for\nparametric bilinear systems, and then provide conditions on projection spaces\nfor the interpolation of structured subsystem transfer functions such that the\nsystem structure and parameter dependencies are preserved in the reduced-order\nmodel. Two benchmark examples with different parameter dependencies are used to\ndemonstrate the theoretical analysis.\n", "doi": "10.1016/j.automatica.2021.109799", "arxiv_id": "2007.11269", "pdf_urls": ["https://arxiv.org/pdf/2007.11269"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica J. IFAC, 132:109799, 2021", "categories": "math.NA cs.NA cs.SY eess.SY math.DS math.OC"} |
| {"title": "Decision-Making in Driver-Automation Shared Control: A Review and\n Perspectives", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Wenshuo Wang", "Xiaoxiang Na", "Dongpu Cao", "Jianwei Gong", "Junqiang Xi", "Yang Xi", "and Fei-Yue Wang"], "abstract": " Shared control schemes allow a human driver to work with an automated driving\nagent in driver-vehicle systems while retaining the driver's abilities to\ncontrol. The human driver, as an essential agent in the driver-vehicle shared\ncontrol systems, should be precisely modeled regarding their cognitive\nprocesses, control strategies, and decision-making processes. The interactive\nstrategy design between drivers and automated driving agents brings an\nexcellent challenge for human-centric driver assistance systems due to the\ninherent characteristics of humans. Many open-ended questions arise, such as\nwhat proper role of human drivers should act in a shared control scheme? How to\nmake an intelligent decision capable of balancing the benefits of agents in\nshared control systems? Due to the advent of these attentions and questions, it\nis desirable to present a survey on the decision-making between human drivers\nand highly automated vehicles, to understand their architectures, human driver\nmodeling, and interaction strategies under the driver-vehicle shared schemes.\nFinally, we give a further discussion on the key future challenges and\nopportunities. They are likely to shape new potential research directions.\ndoi10.1109/jas.2020.1003294arxiv_id2007.12597pdf_urlshttps://arxiv.org/pdf/2007.12597githubis_oasourcesopen-arxivtypejournaljournal_refIEEE/CAA Journal of Automatica Sinica, Vol. 7, No. 5, pp. 1289\n --1307, 2020categorieseess.SY cs.SY |
| titleInvestigating the effect of edge modifications on networked control\n systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsGustav Lindmark and Claudio Altafiniabstract This paper investigates the impact of addition/removal of edges in a complex\nnetworked control system, for the purposes of improving its controllability,\nsystem performances or robustness to external disturbances. The transfer\nfunction formulation we obtain allows to quantify the impact of an edge\nmodification with the $\\hinf$ and $\\htwo$ norms. For stable networks with\npositive edge weights, we show that the $\\hinf$ norm can be computed exactly\nfor each possible single edge modification, as well as the associated stability\nmargin. For the $\\htwo$ norm we instead obtain a lower bound. Since this bound\nis linked to the trace of the controllability Gramian, it can be used for\ninstance to reduce the energy needed for control. When instead the dynamics is\nof Laplacian type, then the norms become unbounded. However, the associated\ndisplacement systems are stable and for them the effect of edge modifications\ncan be quantified. In particular, in this case we provide an upper bound on the\n$ \\hinf$ norm and compute the exact value of the $ \\htwo $ norm for arbitrary\nedge additions.\ndoi10.1016/j.automatica.2022.110801arxiv_id2007.13713pdf_urlshttps://arxiv.org/pdf/2007.13713githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 149:110801, 2023categorieseess.SY cs.SY math.OC |
| titleIterative Pre-Conditioning for Expediting the Gradient-Descent Method:\n The Distributed Linear Least-Squares Problemyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsKushal ChakrabartiNirupam GuptaNikhil Chopraabstract This paper considers the multi-agent linear least-squares problem in a\nserver-agent network. In this problem, the system comprises multiple agents,\neach having a set of local data points, that are connected to a server. The\ngoal for the agents is to compute a linear mathematical model that optimally\nfits the collective data points held by all the agents, without sharing their\nindividual local data points. This goal can be achieved, in principle, using\nthe server-agent variant of the traditional iterative gradient-descent method.\nThe gradient-descent method converges linearly to a solution, and its rate of\nconvergence is lower bounded by the conditioning of the agents' collective data\npoints. If the data points are ill-conditioned, the gradient-descent method may\nrequire a large number of iterations to converge.\n We propose an iterative pre-conditioning technique that mitigates the\ndeleterious effect of the conditioning of data points on the rate of\nconvergence of the gradient-descent method. We rigorously show that the\nresulting pre-conditioned gradient-descent method, with the proposed iterative\npre-conditioning, achieves superlinear convergence when the least-squares\nproblem has a unique solution. In general, the convergence is linear with\nimproved rate of convergence in comparison to the traditional gradient-descent\nmethod and the state-of-the-art accelerated gradient-descent methods. We\nfurther illustrate the improved rate of convergence of our proposed algorithm\nthrough experiments on different real-world least-squares problems in both\nnoise-free and noisy computation environment.\n", "doi": "10.1016/j.automatica.2021.110095", "arxiv_id": "2008.02856", "pdf_urls": ["https://arxiv.org/pdf/2008.02856"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 137, 2022, p110095", "categories": "math.OC cs.LG cs.SY eess.SY stat.ML"} |
| {"title": "Instability Margin Analysis for Parametrized LTI Systems with Application to Repressilator", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Shinji Hara", "Tetsuya Iwasaki", "Yutaka Hori"], "abstract": "This paper is concerned with a robust instability analysis for the single-input-single-output unstable linear time-invariant (LTI) system under dynamic perturbations. The nominal system itself is possibly perturbed by the static gain of the uncertainty, which would be the case when a nonlinear uncertain system is linearized around an equilibrium point. We define the robust instability radius as the smallest $H_\\infty$ norm of the stable linear perturbation that stabilizes the nominal system. There are two main theoretical results: one is on a partial characterization of unperturbed nominal systems for which the robust instability radius can be calculated exactly, and the other is a numerically tractable procedure for calculating the exact robust instability radius for nominal systems parametrized by a perturbation parameter. The results are applied to the repressilator in synthetic biology, where hyperbolic instability of a unique equilibrium guarantees the persistence of oscillation phenomena in the global sense, and the effectiveness of our linear robust instability analysis is confirmed by numerical simulations.", "doi": "10.1016/j.automatica.2021.110047", "arxiv_id": "2008.10983", "pdf_urls": ["https://arxiv.org/pdf/2008.10983"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 136, 110047, 2022", "categories": "eess.SY cs.SY q-bio.QM"} |
| {"title": "A Survey on Negative Transfer", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Wen Zhang", "Lingfei Deng", "Lei Zhang", "Dongrui Wu"], "abstract": " Transfer learning (TL) utilizes data or knowledge from one or more source\ndomains to facilitate the learning in a target domain. It is particularly\nuseful when the target domain has very few or no labeled data, due to\nannotation expense, privacy concerns, etc. Unfortunately, the effectiveness of\nTL is not always guaranteed. Negative transfer (NT), i.e., leveraging source\ndomain data/knowledge undesirably reduces the learning performance in the\ntarget domain, has been a long-standing and challenging problem in TL. Various\napproaches have been proposed in the literature to handle it. However, there\ndoes not exist a systematic survey on the formulation of NT, the factors\nleading to NT, and the algorithms that mitigate NT. This paper fills this gap,\nby first introducing the definition of NT and its factors, then reviewing about\nfifty representative approaches for overcoming NT, according to four\ncategories: secure transfer, domain similarity estimation, distant transfer,\nand NT mitigation. NT in related fields, e.g., multi-task learning, lifelong\nlearning, and adversarial attacks, are also discussed.\n", "doi": "10.1109/jas.2022.106004", "arxiv_id": "2009.00909", "pdf_urls": ["https://arxiv.org/pdf/2009.00909"], "github": ["https://github.com/chamwen/NT-Benchmark"], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "IEEE/CAA Journal of Automatica Sinica, 2022, 1-25", "categories": "cs.LG cs.CV stat.ML"} |
| {"title": "Functional sets with typed symbols : Mixed zonotopes and Polynotopes for\n hybrid nonlinear reachability and filtering", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Christophe Combastel"], "abstract": " Verification and synthesis of Cyber-Physical Systems (CPS) are challenging\nand still raise numerous issues so far. In this paper, based on a new concept\nof mixed sets defined as function images of symbol type domains, a\ncompositional approach combining eager and lazy evaluations is proposed. Syntax\nand semantics are explicitly distinguished. Both continuous (interval) and\ndiscrete (signed, boolean) symbol types are used to model dependencies through\nlinear and polynomial functions, so leading to mixed zonotopic and polynotopic\nsets. Polynotopes extend sparse polynomial zonotopes with typed symbols.\nPolynotopes can both propagate a mixed encoding of intervals and describe the\nbehavior of logic gates. A functional completeness result is given, as well as\nan inclusion method for elementary nonlinear and switching functions. A\nPolynotopic Kalman Filter (PKF) is then proposed as a hybrid nonlinear\nextension of Zonotopic Kalman Filters (ZKF). Bridges with a stochastic\nuncertainty paradigm are briefly outlined. Finally, several discrete,\ncontinuous and hybrid numerical examples including comparisons illustrate the\neffectiveness of the theoretical results.\n", "doi": "10.1016/j.automatica.2022.110457", "arxiv_id": "2009.07387", "pdf_urls": ["https://arxiv.org/pdf/2009.07387"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 143 (2022) 110457", "categories": "eess.SY cs.AI cs.SY math.CO math.OC"} |
| {"title": "Discrimination of attractors with noisy nodes in Boolean networks", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Xiaoqing Cheng", "Wai-Ki Ching", "Sini Guo", "Tatsuya Akutsu"], "abstract": " Observing the internal state of the whole system using a small number of\nsensor nodes is important in analysis of complex networks. Here, we study the\nproblem of determining the minimum number of sensor nodes to discriminate\nattractors under the assumption that each attractor has at most K noisy nodes.\nWe present exact and approximation algorithms for this minimization problem.\nThe effectiveness of the algorithms is also demonstrated by computational\nexperiments using both synthetic data and realistic biological data.\n", "doi": "10.1016/j.automatica.2021.109630", "arxiv_id": "2009.13198", "pdf_urls": ["https://arxiv.org/pdf/2009.13198"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 130 (2021) 109630", "categories": "cs.DS cs.SY eess.SY"} |
| {"title": "Towards a Theoretical Foundation of PID Control for Uncertain Nonlinear\n Systems", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Cheng Zhao and Lei Guo"], "abstract": " As is well-known, the classical PID control plays a dominating role in\nvarious control loops of industrial processes. However, a theory that can\nexplain the rationale why the linear PID can successfully deal with the\nubiquitous uncertain nonlinear dynamical systems and a method that can provide\nexplicit design formulae for the PID parameters are still lacking. This paper\nis a continuation of the authors recent endeavor towards establishing a\ntheoretical foundation of PID. We will investigate the rationale of PID control\nfor a general class of high dimensional second order non-affine uncertain\nsystems. We will show that a three dimensional parameter set can be constructed\nexplicitly, such that whenever the PID parameters are chosen from this set, the\nclosed-loop systems will be globally stable and the regulation error will\nconverge to zero exponentially fast, under some suitable conditions on the\nsystem uncertainties. Moreover, we will show that the PD(PI) control can\nglobally stabilize several special classes of high dimensional uncertain\nnonlinear systems. Furthermore, we will apply the Markus-Yamabe theorem in\ndifferential equations to provide a necessary and sufficient condition for the\nchoice of the PI parameters for a class of one dimensional non-affine uncertain\nsystems. These theoretical results show explicitly that the controller\nparameters are not necessary to be of high gain, and that the ubiquitous PID\ncontrol does indeed have strong robustness with respect to both the system\nstructure uncertainties and the selection of the controller parameters.\n", "doi": "10.1016/j.automatica.2022.110360", "arxiv_id": "2010.06864", "pdf_urls": ["https://arxiv.org/pdf/2010.06864"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2022", "categories": "math.OC"} |
| {"title": "Predicting the future state of disturbed LTI systems: A solution based\n on high-order observers", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Alberto Castillo", "Pedro Garcia"], "abstract": " Predicting the state of a system in a relatively near future time instant is\noften needed for control purposes. However, when the system is affected by\nexternal disturbances, its future state is dependent on the forthcoming\ndisturbance; which is, in most of the cases, unknown and impossible to measure.\nIn this scenario, making predictions of the future system-state is not\nstraightforward and, indeed, there are scarce contributions provided to this\nissue. This paper treats the following problem: given a LTI system affected by\ncontinuously differentiable unknown disturbances, how its future state can be\npredicted in a sufficiently small time-horizon by high-order observers.\nObserver design methodologies in order to reduce the prediction errors are\ngiven. Comparisons with other solutions are also established.\n", "doi": null, "arxiv_id": "2011.01093", "pdf_urls": ["https://arxiv.org/pdf/2011.01093"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 57 (2021), Issue 2 (February)", "categories": "eess.SY cs.SY"} |
| {"title": "Extremum Seeking Control with Attenuated Steady-State Oscillations", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Diganta Bhattacharjee", "Kamesh Subbarao"], "abstract": " We propose two perturbation-based extremum seeking control (ESC) schemes for\ngeneral single input single output nonlinear dynamical systems, having\nstructures similar to that of the classical ESC scheme. We propose novel\nadaptation laws for the excitation signal amplitudes in each scheme that drive\nthe amplitudes to zero. The rates of decay for both the laws are governed by\nthe gradient measures of the unknown reference-to-output equilibrium map. We\nshow that the proposed ESC schemes achieve practical asymptotic convergence to\nthe extremum with a proper tuning of the parameters in the proposed schemes. As\nthe extremum is reached, and the magnitudes of the gradient measures become\nsmall, the excitation signal amplitudes converge to zero. Thus, the proposed\nschemes ensure that the excitation signal is attenuated as the system output\nconverges to a neighborhood of the extremum and the steady-state oscillations\nabout the extremum, typically observed for the classical ESC schemes, are\nattenuated. Simulation examples are included to illustrate the effectiveness of\nthe proposed schemes.\n", "doi": "10.1016/j.automatica.2020.109432", "arxiv_id": "2011.02518", "pdf_urls": ["https://arxiv.org/pdf/2011.02518"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 125, March 2021, 109432", "categories": "eess.SY cs.SY"} |
| {"title": "Online Learning Based Risk-Averse Stochastic MPC of Constrained Linear\n Uncertain Systems", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Chao Ning", "Fengqi You"], "abstract": " This paper investigates the problem of designing data-driven stochastic Model\nPredictive Control (MPC) for linear time-invariant systems under additive\nstochastic disturbance, whose probability distribution is unknown but can be\npartially inferred from data. We propose a novel online learning based\nrisk-averse stochastic MPC framework in which Conditional Value-at-Risk (CVaR)\nconstraints on system states are required to hold for a family of distributions\ncalled an ambiguity set. The ambiguity set is constructed from disturbance data\nby leveraging a Dirichlet process mixture model that is self-adaptive to the\nunderlying data structure and complexity. Specifically, the structural property\nof multimodality is exploit-ed, so that the first- and second-order moment\ninformation of each mixture component is incorporated into the ambiguity set. A\nnovel constraint tightening strategy is then developed based on an equivalent\nreformulation of distributionally ro-bust CVaR constraints over the proposed\nambiguity set. As more data are gathered during the runtime of the controller,\nthe ambiguity set is updated online using real-time disturbance data, which\nenables the risk-averse stochastic MPC to cope with time-varying disturbance\ndistributions. The online variational inference algorithm employed does not\nrequire all collected data be learned from scratch, and therefore the proposed\nMPC is endowed with the guaranteed computational complexity of online learning.\nThe guarantees on recursive feasibility and closed-loop stability of the\nproposed MPC are established via a safe update scheme. Numerical examples are\nused to illustrate the effectiveness and advantages of the proposed MPC.\n", "doi": "10.1016/j.automatica.2020.109402", "arxiv_id": "2011.11441", "pdf_urls": ["https://arxiv.org/pdf/2011.11441"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 125, March 2021, 109402", "categories": "math.OC cs.LG stat.ML"} |
| {"title": "Symbolic Abstractions for Nonlinear Control Systems via Feedback\n Refinement Relation", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Wei Ren and Dimos V. Dimarogonas"], "abstract": " This paper studies the construction of symbolic abstractions for nonlinear\ncontrol systems via feedback refinement relation. Both the delay-free and\ntime-delay cases are addressed. For the delay-free case, to reduce the\ncomputational complexity, we propose a new approximation approach for the state\nand input sets based on a static quantizer, and then a novel symbolic model is\nconstructed such that the original system and the symbolic model satisfy the\nfeedback refinement relation. For the time-delay case, both static and dynamic\nquantizers are combined to approximate the state and input sets. This leads to\na novel dynamic symbolic model for time-delay control systems, and a feedback\nrefinement relation is established between the original system and the symbolic\nmodel. Finally, a numerical example is presented to illustrate the obtained\nresults.\n", "doi": "10.1016/j.automatica.2020.108828", "arxiv_id": "2011.12824", "pdf_urls": ["https://arxiv.org/pdf/2011.12824"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2020", "categories": "eess.SY cs.SY"} |
| {"title": "A Distributed Methodology for Approximate Uniform Global Minimum Sharing", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Michelangelo Bin and Thomas Parisini"], "abstract": " The paper deals with the distributed minimum sharing problem: a set of\ndecision-makers compute the minimum of some local quantities of interest in a\ndistributed and decentralized way by exchanging information through a\ncommunication network. We propose an adjustable approximate solution which\nenjoys several properties of crucial importance in applications. In particular,\nthe proposed solution has good decentralization properties and it is scalable\nin that the number of local variables does not grow with the size or topology\nof the communication network. Moreover, a global and uniform (both in the\ninitial time and in the initial conditions) asymptotic stability result is\nprovided towards a steady state which can be made arbitrarily close to the\nsought minimum. Exact asymptotic convergence can be recovered at the price of\nlosing uniformity with respect to the initial time.\n", "doi": "10.1016/j.automatica.2021.109777", "arxiv_id": "2012.08940", "pdf_urls": ["https://arxiv.org/pdf/2012.08940"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 131, pp. 109777, 2021", "categories": "eess.SY cs.SY"} |
| {"title": "Stabilization of cycles for difference equations with a noisy PF control", "year": 2020, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Elena Braverman", "Josef Dibl\\'ikAlexandra Rodkina and Zden\\v{e}k\n \\v{S}mardaabstract Difference equations, such as a Ricker map, for an increased value of the\nparameter, experience instability of the positive equilibrium and transition to\ndeterministic chaos. To achieve stabilization, various methods can be applied.\nProportional Feedback control suggests a proportional reduction of the state\nvariable at every $k$th step. First, if $k \\neq 1$, a cycle is stabilized\nrather than an equilibrium. Second, the equation can incorporate an additive\nnoise term, describing the variability of the environment, as well as\nmultiplicative noise corresponding to possible deviations in the control\nintensity. The present paper deals with both issues, it justifies a possibility\nof getting a stable blurred $k$-cycle. Presented examples include the Ricker\nmodel, as well as equations with unbounded $f$, such as the bobwhite quail\npopulation models. Though the theoretical results justify stabilization for\neither multiplicative or additive noise only, numerical simulations illustrate\nthat a blurred cycle can be stabilized when both multiplicative and additive\nnoises are involved.\ndoi10.1016/j.automatica.2020.108862arxiv_id2012.11071pdf_urlshttps://arxiv.org/pdf/2012.11071githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica JFAC, V. 115 (2020), paper # 108862, 8 ppcategoriesmath.DS |
| titleMinimal controllability time for systems with nonlinear drift under a\n compact convex state constraintyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsViktor BezborodovLuca Di PersioRiccardo Muradoreabstract In this paper we estimate the minimal controllability time for a class of\nnon-linear control systems with a bounded convex state constraint. An explicit\nexpression is given for the controllability time if the image of the control\nmatrix is of co-dimension one. A lower bound for the controllability time is\ngiven in the general case. The technique is based on finding a lower dimension\nsystem with the similar controllability properties as the original system. The\ncontrols corresponding to the minimal time, or time close to the minimal one,\nare discussed and computed analytically. The effectiveness of the proposed\napproach is illustrated by a few examples.\ndoi10.1016/j.automatica.2020.109428arxiv_id2012.12937pdf_urlshttps://arxiv.org/pdf/2012.12937githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica Volume 125, March 2021,categoriesmath.OC cs.SY eess.SY |
| titleSynchronization with prescribed transient behavior: Heterogeneous\n multi-agent systems under funnel coupling Extended arXiv versionyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsJin Gyu LeeStephan TrennHyungbo Shimabstract In this paper, we introduce a nonlinear time-varying coupling law, which can\nbe designed in a fully decentralized manner and achieves approximate\nsynchronization with arbitrary precision, under only mild assumptions on the\nindividual vector fields and the underlying (undirected) graph structure. The\nproposed coupling law is motivated by the so-called funnel control method\nstudied in adaptive control under the observation that arbitrary precision\nsynchronization can be achieved for heterogeneous multi-agent systems by a\nhigh-gain coupling; consequently we call our novel synchronization method\n`(node-wise) funnel coupling.' By adjusting the conventional proof technique in\nthe funnel control study, we are even able to obtain asymptotic synchronization\nwith the same funnel coupling law. Moreover, the emergent collective behavior\nthat arises for a heterogeneous multi-agent system when enforcing arbitrary\nprecision synchronization by the proposed funnel coupling law, is analyzed in\nthis paper. In particular, we introduce a single scalar dynamics called\n`emergent dynamics' which describes the emergent synchronized behavior of the\nmulti-agent system under funnel coupling. Characterization of the emergent\ndynamics is important because, for instance, one can design the emergent\ndynamics first such that the solution trajectory behaves as desired, and then,\nprovide a design guideline to each agent so that the constructed vector fields\nyield the desired emergent dynamics. We illustrate this idea via the example of\na distributed median solver based on funnel coupling.\ndoi10.1016/j.automatica.2022.110276arxiv_id2012.14580pdf_urlshttps://arxiv.org/pdf/2012.14580githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 141, pp. 110276, July 2022categorieseess.SY cs.SY |
| titleOptimal adaptive testing for epidemic control: combining molecular and\n serology testsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsD. AcemogluA. FallahA. GiomettoD. HuttenlocherA. OzdaglarF.\n PariseS. Pattathilabstract The COVID-19 crisis highlighted the importance of non-medical interventions,\nsuch as testing and isolation of infected individuals, in the control of\nepidemics. Here, we show how to minimize testing needs while maintaining the\nnumber of infected individuals below a desired threshold. We find that the\noptimal policy is adaptive, with testing rates that depend on the epidemic\nstate. Additionally, we show that such epidemic state is difficult to infer\nwith molecular tests alone, which are highly sensitive but have a short\ndetectability window. Instead, we propose the use of baseline serology testing,\nwhich is less sensitive but detects past infections, for the purpose of state\nestimation. Validation of such combined testing approach with a stochastic\nmodel of epidemics shows significant cost savings compared to non-adaptive\ntesting strategies that are the current standard for COVID-19.\ndoi10.1016/j.automatica.2023.111391arxiv_id2101.00773pdf_urlshttps://arxiv.org/pdf/2101.00773githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 160, February 2024, 111391categorieseess.SY cs.SY math.DS math.OC physics.soc-ph q-bio.PE |
| titleAn Analytic Layer-wise Deep Learning Framework with Applications to\n Roboticsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsHuu-Thiet NguyenChien Chern CheahKar-Ann Tohabstract Deep learning (DL) has achieved great success in many applications, but it\nhas been less well analyzed from the theoretical perspective. The unexplainable\nsuccess of black-box DL models has raised questions among scientists and\npromoted the emergence of the field of explainable artificial intelligence\n(XAI). In robotics, it is particularly important to deploy DL algorithms in a\npredictable and stable manner as robots are active agents that need to interact\nsafely with the physical world. This paper presents an analytic deep learning\nframework for fully connected neural networks, which can be applied for both\nregression problems and classification problems. Examples for regression and\nclassification problems include online robot control and robot vision. We\npresent two layer-wise learning algorithms such that the convergence of the\nlearning systems can be analyzed. Firstly, an inverse layer-wise learning\nalgorithm for multilayer networks with convergence analysis for each layer is\npresented to understand the problems of layer-wise deep learning. Secondly, a\nforward progressive learning algorithm where the deep networks are built\nprogressively by using single hidden layer networks is developed to achieve\nbetter accuracy. It is shown that the progressive learning method can be used\nfor fine-tuning of weights from convergence point of view. The effectiveness of\nthe proposed framework is illustrated based on classical benchmark recognition\ntasks using the MNIST and CIFAR-10 datasets and the results show a good balance\nbetween performance and explainability. The proposed method is subsequently\napplied for online learning of robot kinematics and experimental results on\nkinematic control of UR5e robot with unknown model are presented.\ndoi10.1016/j.automatica.2021.110007arxiv_id2102.03705pdf_urlshttps://arxiv.org/pdf/2102.03705githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 135, Jan. 2022categoriescs.RO cs.SY eess.SY |
| titleInterval Analysis of Worst-case Stationary Moments for Stochastic\n Chemical Reactions with Uncertain Parametersyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsYuta SakuraiYutaka Horiabstract The dynamics of cellular chemical reactions are variable due to stochastic\nnoise from intrinsic and extrinsic sources. The intrinsic noise is the\nintracellular fluctuations of molecular copy numbers caused by the\nprobabilistic encounter of molecules and is modeled by the chemical master\nequation. The extrinsic noise, on the other hand, represents the intercellular\nvariation of the kinetic parameters due to the variation of global factors\naffecting gene expression. The objective of this paper is to propose a\ntheoretical framework to analyze the combined effect of the intrinsic and the\nextrinsic noise modeled by the chemical master equation with uncertain\nparameters. More specifically, we formulate a semidefinite program to compute\nthe intervals of the stationary solution of uncertain moment equations whose\nparameters are given only partially in the form of the statistics of their\ndistributions. The semidefinite program is derived without approximating the\ngoverning equation in contrast with many existing approaches. Thus, we can\nobtain guaranteed intervals of the worst possible values of the moments for all\nparameter distributions satisfying the given statistics, which are\nprohibitively hard to estimate from sample-path simulations since sampling from\nall possible uncertain distributions is difficult. We demonstrate the proposed\noptimization approach using two examples of stochastic chemical reactions and\nshow that the solution of the optimization problem gives informative upper and\nlower bounds of the statistics of the stationary copy number distributions.\ndoi10.1016/j.automatica.2022.110647arxiv_id2102.04237pdf_urlshttps://arxiv.org/pdf/2102.04237githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 110647, 2022categorieseess.SY cs.SY q-bio.QM |
| titleAnts and bracket generating distributions in dimension 5 and 6yearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAndrei Agrachev and Pawel Nurowskiabstract We consider a mechanical system of three ants on the floor, which move\naccording to two independt rules: Rule A - forces the velocity of any given ant\nto always point at a neighboring ant, and Rule B - forces the velocity of every\nant to be parallel to the line defined by the two other ants. We observe that\nRule A equips the 6-dimensional configuration space of the ants with a\nstructure of a homogeneous (3,6) distribution, and that Rule B foliates this\n6-dimensional configuration space onto 5-dimensional leaves, each of which is\nequiped with a homogeneous (2,3,5) distribution. The symmetry properties and\nBryant-Cartan local invariants of these distributions are determined.\n In the case of Rule B we study and determine the singular trajectories\n(abnormal extremals) of the corresponding distributions. We show that these\nsatisfy an interesting system of two ODEs of Fuchsian type.\ndoi10.1016/j.automatica.2022.110693arxiv_id2103.01058pdf_urlshttps://arxiv.org/pdf/2103.01058githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica 147, (2023), 110693categoriesmath.DG math.OC |
| titleStability analysis of reaction-diffusion PDEs coupled at the boundaries\n with an ODEyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsHugo Lhachemi and Christophe Prieurabstract This paper addresses the derivation of generic and tractable sufficient\nconditions ensuring the stability of a coupled system composed of a\nreaction-diffusion partial differential equation (PDE) and a finite-dimensional\nlinear time invariant ordinary differential equation (ODE). The coupling of the\nPDE with the ODE is located either at the boundaries or in the domain of the\nreaction-diffusion equation and takes the form of the input and output of the\nODE. We investigate boundary Dirichlet/Neumann/Robin couplings, as well as\nin-domain Dirichlet/Neumann couplings. The adopted approach relies on the\nspectral reduction of the problem by projecting the trajectory of the PDE into\na Hilbert basis composed of the eigenvectors of the underlying Sturm-Liouville\noperator and yields a set of sufficient stability conditions taking the form of\nLMIs. We propose numerical examples, consisting of an unstable\nreaction-diffusion equation and an unstable ODE, such that the application of\nthe derived stability conditions ensure the stability of the resulting coupled\nPDE-ODE system.\ndoi10.1016/j.automatica.2022.110465arxiv_id2103.04775pdf_urlshttps://arxiv.org/pdf/2103.04775githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 144, October 2022, 110465categoriesmath.OC cs.SY eess.SY |
| titleOn Asymptotic Log-Optimal Buy-and-Hold Strategyyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsChung-Han Hsiehabstract In this paper, we consider a frequency-based portfolio optimization problem\nwith $m \\geq 2$ assets when the expected logarithmic growth (ELG) rate of\nwealth is used as the performance metric. With the aid of the notion called\ndominant asset, it is known that the optimal ELG level is achieved by investing\nall available funds on that asset. However, such an \ strategy is\narguably too risky to implement in practice. Motivated by this issue, we study\nthe case where the portfolio weights are chosen in a rather ad-hoc manner and a\nbuy-and-hold strategy is subsequently used. Then we show that, if the\nunderlying portfolio contains a dominant asset, buy and hold on that specific\nasset is asymptotically log-optimal with a sublinear rate of convergence. This\nresult also extends to the scenario where a trader either does not have a\nprobabilistic model for the returns or does not trust a model obtained from\nhistorical data. To be more specific, we show that if a market contains a\ndominant asset, buy and hold a market portfolio involving nonzero weights for\neach asset is asymptotically log-optimal. Additionally, this paper also\nincludes a conjecture regarding the property called high-frequency maximality.\nThat is, in the absence of transaction costs, high-frequency rebalancing is\nunbeatable in the ELG sense. Support for the conjecture, involving a lemma for\na weak version of the conjecture, is provided. This conjecture, if true,\nenables us to improve the log-optimality result obtained previously. Finally, a\nresult that indicates a way regarding an issue about when should one to\nrebalance their portfolio if needed, is also provided. Examples, some involving\nsimulations with historical data, are also provided along the way to illustrate\nthe~theory.\ndoi10.1016/j.automatica.2023.110901arxiv_id2103.04898pdf_urlshttps://arxiv.org/pdf/2103.04898githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 151, pp. 110901:1-110901:11, 2023categoriesq-fin.PM q-fin.CP q-fin.MF |
| titleControl Design with Guaranteed Transient Performance: an Approach with\n Polyhedral Target Tubesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsWillem EsterhuizenQing-Guo Wangabstract In this paper a novel approach is presented for control design with\nguaranteed transient performance for multiple-input multiple-output\ndiscrete-time linear polytopic difference inclusions. We establish a theorem\nthat gives necessary and sufficient conditions for the state to evolve from one\npolyhedral subset of the state-space to another. Then we present an algorithm\nwhich constructs a time-varying output feedback law which guarantees that the\nstate evolves within a time-varying polyhedral target-tube specifying the\nsystem's desired transient performance. We present generalisations involving\nconstraints on the control, and a bounded additive disturbance term. Our\nformulation is very general and includes reference tracking with any desired\ntransient behaviour in the face of disturbances, as specified, for example, by\nthe most popular step response specifications. The approach is demonstrated by\nan example involving the control of water levels in two coupled tanks.\n", "doi": "10.1016/j.automatica.2020.109097", "arxiv_id": "2103.05333", "pdf_urls": ["https://arxiv.org/pdf/2103.05333"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 119, 109097 (2020)", "categories": "math.OC cs.SY eess.SY"} |
| {"title": "Bayesian adversarial multi-node bandit for optimal smart grid protection\n against cyber attacks", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Jianyu Xu", "Bin Liu", "Huadong Mo", "Daoyi Dong"], "abstract": " The cybersecurity of smart grids has become one of key problems in developing\nreliable modern power and energy systems. This paper introduces a\nnon-stationary adversarial cost with a variation constraint for smart grids and\nenables us to investigate the problem of optimal smart grid protection against\ncyber attacks in a relatively practical scenario. In particular, a Bayesian\nmulti-node bandit (MNB) model with adversarial costs is constructed and a new\nregret function is defined for this model. An algorithm called Thompson-Hedge\nalgorithm is presented to solve the problem and the superior performance of the\nproposed algorithm is proven in terms of the convergence rate of the regret\nfunction. The applicability of the algorithm to real smart grid scenarios is\nverified and the performance of the algorithm is also demonstrated by numerical\nexamples.\n", "doi": null, "arxiv_id": "2104.02774", "pdf_urls": ["https://arxiv.org/pdf/2104.02774"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 2021", "categories": "cs.CR cs.LG cs.SY eess.SY"} |
| {"title": "Safe Exploration in Model-based Reinforcement Learning using Control\n Barrier Functions", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Max H. Cohen and Calin Belta"], "abstract": " This paper develops a model-based reinforcement learning (MBRL) framework for\nlearning online the value function of an infinite-horizon optimal control\nproblem while obeying safety constraints expressed as control barrier functions\n(CBFs). Our approach is facilitated by the development of a novel class of\nCBFs, termed Lyapunov-like CBFs (LCBFs), that retain the beneficial properties\nof CBFs for developing minimally-invasive safe control policies while also\npossessing desirable Lyapunov-like qualities such as positive\nsemi-definiteness. We show how these LCBFs can be used to augment a\nlearning-based control policy to guarantee safety and then leverage this\napproach to develop a safe exploration framework in a MBRL setting. We\ndemonstrate that our approach can handle more general safety constraints than\ncomparative methods via numerical examples.\n", "doi": "10.1016/j.automatica.2022.110684", "arxiv_id": "2104.08171", "pdf_urls": ["https://arxiv.org/pdf/2104.08171"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 147, p. 110684, 2023", "categories": "cs.LG cs.SY eess.SY"} |
| {"title": "Integral Line-of-Sight Path Following Control of Magnetic Helical\n Microswimmers Subject to Step-Out Frequencies", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Alireza Mohammadi", "Mark W. Spong"], "abstract": " This paper investigates the problem of straight-line path following for\nmagnetic helical microswimmers. The control objective is to make the helical\nmicroswimmer to converge to a straight line without violating the step-out\nfrequency constraint. The proposed feedback control solution is based on an\noptimal decision strategy (ODS) that is cast as a trust-region subproblem\n(TRS), i.e., a quadratic program over a sphere. The ODS-based control strategy\nminimizes the difference between the microrobot velocity and an integral\nline-of-sight (ILOS)-based reference vector field while respecting the magnetic\nsaturation constraints and ensuring the absolute continuity of the control\ninput. Due to the embedded integral action in the reference vector field, the\nmicroswimmer will follow the desired straight line by compensating for the\ndrift effect of the environmental disturbances as well as the microswimmer\nweight.\n", "doi": "10.1016/j.automatica.2021.109554", "arxiv_id": "2104.14241", "pdf_urls": ["https://arxiv.org/pdf/2104.14241"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 128, June 2021, 109554", "categories": "eess.SY cs.RO cs.SY math.OC"} |
| {"title": "Identity Concealment Games: How I Learned to Stop Revealing and Love the\n Coincidences", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Mustafa O. Karabag", "Melkior Ornik", "Ufuk Topcu"], "abstract": " In an adversarial environment, a hostile player performing a task may behave\nlike a non-hostile one in order not to reveal its identity to an opponent. To\nmodel such a scenario, we define identity concealment games: zero-sum\nstochastic reachability games with a zero-sum objective of identity\nconcealment. To measure the identity concealment of the player, we introduce\nthe notion of an average player. The average player's policy represents the\nexpected behavior of a non-hostile player. We show that there exists an\nequilibrium policy pair for every identity concealment game and give the\noptimality equations to synthesize an equilibrium policy pair. If the player's\nopponent follows a non-equilibrium policy, the player can hide its identity\nbetter. For this reason, we study how the hostile player may learn the\nopponent's policy. Since learning via exploration policies would quickly reveal\nthe hostile player's identity to the opponent, we consider the problem of\nlearning a near-optimal policy for the hostile player using the game runs\ncollected under the average player's policy. Consequently, we propose an\nalgorithm that provably learns a near-optimal policy and give an upper bound on\nthe number of sample runs to be collected.\ndoiarxiv_id2105.05377pdf_urlshttps://arxiv.org/pdf/2105.05377githubis_oasourcesopen-arxivtypejournaljournal_refKarabag, Mustafa O., Melkior Ornik, and Ufuk Topcu. \ Automatica 161 (2024): 111482categoriescs.GT cs.MA math.OC |
| titlePredictor-Based Output Feedback Stabilization of an Input Delayed\n Parabolic PDE with Boundary Measurementyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsHugo Lhachemi and Christophe Prieurabstract This paper is concerned with the output feedback boundary stabilization of\ngeneral 1-D reaction diffusion PDEs in the presence of an arbitrarily large\ninput delay. We consider the cases of Dirichlet/Neumann/Robin boundary\nconditions for the both boundary control and boundary condition. The boundary\nmeasurement takes the form of a either Dirichlet or Neumann trace. The adopted\ncontrol strategy is composed of a finite-dimensional observer estimating the\nfirst modes of the PDE coupled with a predictor to compensate the input delay.\nIn this context, we show for any arbitrary value of the input delay that the\ncontrol strategy achieves the exponential stabilization of the closed-loop\nsystem, for system trajectories evaluated in $H^1$ norm (also in $L^2$ norm in\nthe case of a Dirichlet boundary measurement), provided the dimension of the\nobserver is selected large enough. The reported proof of this result requires\nto perform both control design and stability analysis using simultaneously the\n(non-homogeneous) original version of the PDE and one of its equivalent\nhomogeneous representations.\ndoi10.1016/j.automatica.2021.110115arxiv_id2105.08431pdf_urlshttps://arxiv.org/pdf/2105.08431githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 137, March 2022, 110115categoriesmath.OC cs.SY eess.SY |
| titleDistributed adaptive stabilizationyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsZhiyong SunAnders RantzerZhongkui LiAnders Robertssonabstract In this paper we consider distributed adaptive stabilization for uncertain\nmultivariable linear systems with a time-varying diagonal matrix gain. We show\nthat uncertain multivariable linear systems are stabilizable by diagonal matrix\nhigh gains if the system matrix is an H-matrix with positive diagonal entries.\nBased on matrix measure and stability theory for diagonally dominant systems,\nwe consider two classes of uncertain linear systems, and derive a threshold\ncondition to ensure their exponential stability by a monotonically increasing\ndiagonal gain matrix. When each individual gain function in the matrix gain is\nupdated by state-dependent functions using only local state information, the\nboundedness and convergence of both system states and adaptive matrix gains are\nguaranteed. We apply the adaptive distributed stabilization approach to\nadaptive synchronization control for large-scale complex networks consisting of\nnonlinear node dynamics and time-varying coupling weights. A unified framework\nfor adaptive synchronization is proposed that includes several general design\napproaches for adaptive coupling weights to guarantee network synchronization.\ndoi10.1016/j.automatica.2021.109616arxiv_id2105.14004pdf_urlshttps://arxiv.org/pdf/2105.14004githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica: Volume 129, 109616 (1-13), July 2021categorieseess.SY cs.DC cs.MA cs.SY math.OC nlin.AO |
| titleAn arbitrary-order predefined-time exact differentiator for signals with\n exponential growth boundyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsDavid G\\'omez-Guti\\'errezRodrigo Aldana-L\\'opez", "Richard Seeber", "Marco Tulio Angulo", "Leonid Fridman"], "abstract": " There is a growing interest in differentiation algorithms that converge in\nfixed time with a predefined Upper Bound on the Settling Time (UBST). However,\nexisting differentiation algorithms are limited to signals having an $n$-th\norder Lipschitz derivative. Here, we introduce a general methodology based on\ntime-varying gains to circumvent this limitation, allowing us to design $n$-th\norder differentiators with a predefined UBST for the broader class of signals\nwhose $(n+1)$-th derivative is bounded by a function with bounded logarithmic\nderivative. Unlike existing methods whose time-varying gain tends to infinity,\nour approach yields a time-varying gain that remains bounded at convergence\ntime. We show how this last property maintains exact convergence using bounded\ngains when considering a compact set of initial conditions and improves the\nalgorithm's performance to measurement noise.\ndoi10.1016/j.automatica.2023.110995.arxiv_id2106.00822pdf_urlshttps://arxiv.org/pdf/2106.00822githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2023categoriesmath.OC cs.SY eess.SY |
| titleClosed-form H-infinity Optimal Control for a Class of\n Infinite-Dimensional Systemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsCarolina BergelingKirsten A. MorrisAnders Rantzerabstract H-infinity optimal control and estimation are addressed for a class of\nsystems governed by partial differential equations with bounded input and\noutput operators. Diffusion equations are an important example in this class.\nExplicit formulas for the optimal state feedback controller as well as the\noptimal state estimator are given. Unlike traditional methods for H-infinity\nsynthesis, no iteration is needed to obtain the optimal solution. Moreover, the\noptimal performance for both the state feedback and state estimation problems\nare explicitly calculated. This is shown to be useful for problems of\nH-infinity optimal actuator and sensor location. Furthermore, the results can\nbe used in testing and bench-marking of general purpose algorithms for\nH-infinity synthesis. The results also apply to finite-dimensional systems.\ndoi10.1016/j.automatica.2020.108916arxiv_id2106.04183pdf_urlshttps://arxiv.org/pdf/2106.04183githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 117, 2020, 108916, ISSN 0005-1098categoriesmath.OC |
| titleRevealing the canalizing structure of Boolean functions: Algorithms and\n applicationsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsElena DimitrovaBrandilyn StiglerClaus KadelkaDavid Murrugarraabstract Boolean functions can be represented in many ways including logical forms,\ntruth tables, and polynomials. Additionally, Boolean functions have different\ncanonical representations such as minimal disjunctive normal forms. Other\ncanonical representation is based on the polynomial representation of Boolean\nfunctions where they can be written as a nested product of canalizing layers\nand a polynomial that contains the noncanalizing variables. In this paper we\nstudy the problem of identifying the canalizing layers format of Boolean\nfunctions. First, we show that the problem of finding the canalizing layers is\nNP-hard. Second, we present several algorithms for finding the canalizing\nlayers of a Boolean function, discuss their complexities, and compare their\nperformances. Third, we show applications where the computation of canalizing\nlayers can be used for finding a disjunctive normal form of a nested canalizing\nfunction. Another application deals with the reverse engineering of Boolean\nnetworks with a prescribed layering format. Finally, implementations of our\nalgorithms in Python and in the computer algebra system Macaulay2 are available\nat https://github.com/ckadelka/BooleanCanalization.\ndoi10.1016/j.automatica.2022.110630arxiv_id2106.06580pdf_urlshttps://arxiv.org/pdf/2106.06580githubhttps://github.com/ckadelka/BooleanCanalizationis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 146 (2022)categoriescs.CC q-bio.QM |
| titleGlobal Structure Identifiability and Reconstructibility of an NDS with\n Descriptor Subsystemsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsTong Zhou and Kailin Yinabstract This paper investigates requirements on a networked dynamic system (NDS) such\nthat its subsystem interactions can be solely determined from experiment data\nor reconstructed from its overall model. The NDS is constituted from several\nsubsystems whose dynamics are described through a descriptor form. Except\nregularity on each subsystem and the whole NDS, no other restrictions are put\non either subsystem dynamics or subsystem interactions. A matrix rank based\nnecessary and sufficient condition is derived for the global identifiability of\nsubsystem interactions, which leads to several conclusions about NDS structure\nidentifiability when there is some a priori information. This matrix also gives\nan explicit description for the set of subsystem interactions that can not be\ndistinguished from experiment data only. In addition, under a well-posedness\nassumption, a necessary and sufficient condition is obtained for the\nreconstructibility of subsystem interactions from an NDS descriptor form model.\nThis condition can be verified with each subsystem separately and is therefore\nattractive in the analysis and synthesis of a large-scale NDS. Simulation\nresults show that rather than increases monotonically with the distance of\nsubsystem interactions to the undifferentiable set, the magnitude of the\nexternal output differences between two NDSs with distinct subsystem\ninteractions increases much more rapidly when one of them is close to be\nunstable. In addition, directions of probing signals are also very important in\ndistinguishing external outputs of distinctive NDSs.These findings are expected\nto be helpful in identification experiment designs, etc.\ndoi10.1016/j.automatica.2022.110356arxiv_id2107.01124pdf_urlshttps://arxiv.org/pdf/2107.01124githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Vol.142, 110356, 2022categorieseess.SY cs.SY math.OC |
| titleTrajectory Planning Under Environmental Uncertainty With Finite-Sample\n Safety Guaranteesyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsVasileios LefkopoulosMaryam Kamgarpourabstract We tackle the problem of trajectory planning in an environment comprised of a\nset of obstacles with uncertain time-varying locations. The uncertainties are\nmodeled using widely accepted Gaussian distributions, resulting in a\nchance-constrained program. Contrary to previous approaches however, we do not\nassume perfect knowledge of the moments of the distribution, and instead\nestimate them through finite samples available from either sensors or past\ndata. We derive tight concentration bounds on the error of these estimates to\nsufficiently tighten the chance-constraint program. As such, we provide\nprovable guarantees on satisfaction of the chance-constraints corresponding to\nthe nominal yet unknown moments. We illustrate our results with two autonomous\nvehicle trajectory planning case studies.\ndoi10.1016/j.automatica.2021.109754arxiv_id2108.06250pdf_urlshttps://arxiv.org/pdf/2108.06250githubhttps://github.com/exomag/finite-sample-uncertainty-planning-automaticais_oasourcesopen-arxivtypejournaljournal_refAutomatica Volume 131, 2021, Page(s) 109754categorieseess.SY cs.SY math.OC |
| titleRobust Tube-based Model Predictive Control with Koopman\n Operators--Extended Versionyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsXinglong ZhangWei PanRiccardo ScattoliniShuyou Yuand Xin Xuabstract Koopman operators are of infinite dimension and capture the characteristics\nof nonlinear dynamics in a lifted global linear manner. The finite data-driven\napproximation of Koopman operators results in a class of linear predictors,\nuseful for formulating linear model predictive control (MPC) of nonlinear\ndynamical systems with reduced computational complexity. However, the\nrobustness of the closed-loop Koopman MPC under modeling approximation errors\nand possible exogenous disturbances is still a crucial issue to be resolved.\nAiming at the above problem, this paper presents a robust tube-based MPC\nsolution with Koopman operators, i.e., r-KMPC, for nonlinear discrete-time\ndynamical systems with additive disturbances. The proposed controller is\ncomposed of a nominal MPC using a lifted Koopman model and an off-line\nnonlinear feedback policy. The proposed approach does not assume the\nconvergence of the approximated Koopman operator, which allows using a Koopman\nmodel with a limited order for controller design. Fundamental properties, e.g.,\nstabilizability, observability, of the Koopman model are derived under standard\nassumptions with which, the closed-loop robustness and nominal point-wise\nconvergence are proven. Simulated examples are illustrated to verify the\neffectiveness of the proposed approach.\ndoi10.1016/j.automatica.2021.110114arxiv_id2108.13011pdf_urlshttps://arxiv.org/pdf/2108.13011githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, March, 2022categorieseess.SY cs.SY |
| titleSparse optimal stochastic controlyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsKaito ItoTakuya IkedaKenji Kashimaabstract In this paper, we investigate a sparse optimal control of continuous-time\nstochastic systems. We adopt the dynamic programming approach and analyze the\noptimal control via the value function. Due to the non-smoothness of the $L^0$\ncost functional, in general, the value function is not differentiable in the\ndomain. Then, we characterize the value function as a viscosity solution to the\nassociated Hamilton-Jacobi-Bellman (HJB) equation. Based on the result, we\nderive a necessary and sufficient condition for the $L^0$ optimality, which\nimmediately gives the optimal feedback map. Especially for control-affine\nsystems, we consider the relationship with $L^1$ optimal control problem and\nshow an equivalence theorem.\ndoi10.1016/j.automatica.2020.109438arxiv_id2109.07716pdf_urlshttps://arxiv.org/pdf/2109.07716githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, 2021categoriesmath.OC cs.SY eess.SY |
| titleBarriers and Potentially Safe Sets in Hybrid Systems: Pendulum with\n Non-Rigid Cableyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsWillem Esterhuizen and Jean L\\'evine"], "abstract": " This paper deals with an application of the notion of barrier in mixed\nconstrained nonlinear systems to an example of a pendulum mounted on a cart\nwith non-rigid cable, whose dynamics may switch to free-fall when the tension\nof the cable vanishes. We present a direct construction of the boundary of the\npotentially safe set in which there always exists a control such that the cable\nnever goes slack. A discussion on the dependence of this set with respect to\nthe pendulum and cart masses is then sketched.\n", "doi": "10.1016/j.automatica.2016.07.001", "arxiv_id": "2109.10976", "pdf_urls": ["https://arxiv.org/pdf/2109.10976"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, 73, pp.248-255 (2016)", "categories": "math.OC cs.SY eess.SY"} |
| {"title": "Predictability and Fairness in Load Aggregation and Operations of\n Virtual Power Plants", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Jakub Marecek", "Michal Roubalik", "Ramen Ghosh", "Robert N. Shorten", "Fabian\n R. Wirth"], "abstract": " In power systems, one wishes to regulate the aggregate demand of an ensemble\nof distributed energy resources (DERs), such as controllable loads and battery\nenergy storage systems. We suggest a notion of predictability and fairness,\nwhich suggests that the long-term averages of prices or incentives offered\nshould be independent of the initial states of the operators of the DER, the\naggregator, and the power grid. We show that this notion cannot be guaranteed\nwith many traditional controllers used by the load aggregator, including the\nusual proportional-integral (PI) controller. We show that even considering the\nnon-linearity of the alternating-current model, this notion of predictability\nand fairness can be guaranteed for incrementally input-to-state stable (iISS)\ncontrollers, under mild assumptions.\n", "doi": "10.1016/j.automatica.2022.110743", "arxiv_id": "2110.03001", "pdf_urls": ["https://arxiv.org/pdf/2110.03001"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 147, January 2023, 110743", "categories": "math.OC cs.AI cs.SY eess.SY"} |
| {"title": "Edge-wise funnel output synchronization of heterogeneous agents with\n relative degree one", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Jin Gyu Lee", "Thomas Berger", "Stephan Trenn", "Hyungbo Shim"], "abstract": " When a group of heterogeneous node dynamics are diffusively coupled with a\nhigh coupling gain, the group exhibits a collective emergent behavior which is\ngoverned by a simple algebraic average of the node dynamics called the blended\ndynamics. This finding has been utilized for designing heterogeneous\nmulti-agent systems by building the desired blended dynamics first and then\nsplitting it into the node dynamics. However, to compute the magnitude of the\ncoupling gain, each agent needs to know global information such as the number\nof participating nodes, the graph structure, and so on, which prevents a fully\ndecentralized design of the node dynamics in conjunction with the coupling\nlaws. To resolve this issue, the idea of funnel control, which is a method for\nadaptive gain selection, can be exploited for a node-wise coupling, but the\nprice to pay is that the collective emergent behavior is no longer governed by\na simple average of the node dynamics. Our analysis reveals that this drawback\ncan be avoided by an edge-wise design premise, which is the idea that we\npresent in this paper. After all, we gain benefits such as a fully\ndecentralized design without global information, collective emergent behavior\nbeing governed by the blended dynamics, and the plug-and-play operation based\non edge-wise handshaking between two nodes.\n", "doi": "10.1016/j.automatica.2023.111204", "arxiv_id": "2110.05330", "pdf_urls": ["https://arxiv.org/pdf/2110.05330"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, vol. 156, pp. 111204, Oct 2023", "categories": "eess.SY cs.SY"} |
| {"title": "Formation control of a leader-follower structure in three dimensional\n space using bearing measurements", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Zhiqi Tang", "Rita Cunha", "Tarek Hamel", "Carlos Silvestre"], "abstract": " This paper addresses the problem of bearing leader-follower formation control\nin three-dimensional space by exploring the persistence of excitation (PE) of\nthe desired formation. Using only bearing and relative velocity measurements,\ndistributed control laws are derived for a group of agents with\ndouble-integrator dynamics. The key contribution is that the exponential\nstabilization of the actual formation to the desired one in terms of both shape\nand scale is guaranteed as long as the PE conditions on the desired formation\nare satisfied. The approach generalizes stability results provided in prior\nwork for leader-first follower (LFF) structures which are based on bearing\nrigidity and constraint consistency to ensure the exponential stabilization of\nthe actual formation to a desired static geometric pattern up to a scale\nfactor. Simulations results are provided to illustrate the performance of the\nproposed control method.\n", "doi": null, "arxiv_id": "2110.08054", "pdf_urls": ["https://arxiv.org/pdf/2110.08054"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 128, June 2021, 109567", "categories": "eess.SY cs.SY"} |
| {"title": "The $k$-Compound of a Difference-Algebraic System", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Ron Ofir and Michael Margaliot"], "abstract": "The multiplicative and additive compounds of a matrix have important applications in geometry, linear algebra, and the analysis of dynamical systems. In particular, the $k$-compounds allow to build a $k$-compound dynamical system that tracks the evolution of $k$-dimensional parallelotopes along the original dynamics. This has recently found many applications in the analysis of non-linear systems described by ODEs and difference equations. Here, we introduce the $k$-compound system corresponding to a differential-algebraic system, and describe several applications to the analysis of discrete-time dynamical systems described by difference-algebraic equations.", "doi": "10.1016/j.automatica.2023.111387", "arxiv_id": "2111.01419", "pdf_urls": ["https://arxiv.org/pdf/2111.01419"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica 159 (2024)", "categories": "eess.SY cs.SY"} |
| {"title": "Integral state-feedback control of linear time-varying systems: A\n performance preserving approach", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Richard Seeber and Markus Tranninger"], "abstract": " An integral extension of state-feedback controllers for linear time-varying\nplants is proposed, which preserves performance of the nominal controller in\nthe unperturbed case. Similar to time-invariant state feedback with integral\naction, the controller achieves complete rejection of disturbances whose\neffective action on the plant is constant with respect to the control input.\nMoreover, bounded-input bounded-state stability with respect to arbitrary\ndisturbances is shown. A modification for preventing controller windup as well\nas tuning guidelines are discussed. The efficacy of the proposed technique is\ndemonstrated in simulation for a two-tank system that is linearized along a\ntime-varying reference trajectory.\n", "doi": "10.1016/j.automatica.2021.110000", "arxiv_id": "2111.03380", "pdf_urls": ["https://arxiv.org/pdf/2111.03380"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, volume 136, February 2022, article 110000", "categories": "eess.SY cs.SY"} |
| {"title": "Optimal Robust Exact Differentiation via Linear Adaptive Techniques", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Richard Seeber and Hernan Haimovich"], "abstract": " The problem of differentiating a function with bounded second derivative in\nthe presence of bounded measurement noise is considered in both continuous-time\nand sampled-data settings. Fundamental performance limitations of causal\ndifferentiators, in terms of the smallest achievable worst-case differentiation\nerror, are shown. A robust exact differentiator is then constructed via the\nadaptation of a single parameter of a linear differentiator. It is demonstrated\nthat the resulting differentiator exhibits a combination of properties that\noutperforms existing continuous-time differentiators: it is robust with respect\nto noise, it instantaneously converges to the exact derivative in the absence\nof noise, and it attains the smallest possible -- hence optimal -- upper bound\non its differentiation error under noisy measurements. For sample-based\ndifferentiators, the concept of quasi-exactness is introduced to classify\ndifferentiators that achieve the lowest possible worst-case error based on\nsampled measurements in the absence of noise. A straightforward sample-based\nimplementation of the proposed linear adaptive continuous-time differentiator\nis shown to achieve quasi-exactness after a single sampling step as well as a\ntheoretically optimal differentiation error bound that, in addition, converges\nto the continuous-time optimal one as the sampling period becomes arbitrarily\nsmall. A numerical simulation illustrates the presented formal results.\n", "doi": "10.1016/j.automatica.2022.110725", "arxiv_id": "2111.12638", "pdf_urls": ["https://arxiv.org/pdf/2111.12638"], "github": [], "is_oa": true, "sources": ["open-arxiv"], "type": "journal", "journal_ref": "Automatica, Volume 148, February 2023, Article 110725", "categories": "eess.SY cs.SY"} |
| {"title": "Coordinated Multi-Robot Trajectory Tracking Control over Sampled\n Communication", "year": 2021, "venue_abbr": "Automatica", "venue_name": "Automatica", "authors": ["Enrica Rossi", "Marco Tognon", "Luca Ballotta", "Ruggero Carli", "Juan\n Cort\\'esAntonio FranchiLuca Schenatoabstract In this paper, we propose an inverse-kinematics controller for a class of\nmulti-robot systems in the scenario of sampled communication. The goal is to\nmake a group of robots perform trajectory tracking in a coordinated way when\nthe sampling time of communications is much larger than the sampling time of\nlow-level controllers, disrupting theoretical convergence guarantees of\nstandard control design in continuous time. Given a desired trajectory in\nconfiguration space which is precomputed offline, the proposed controller\nreceives configuration measurements, possibly via wireless, to re-compute\nvelocity references for the robots, which are tracked by a low-level\ncontroller. We propose joint design of a sampled proportional feedback plus a\nnovel continuous-time feedforward that linearizes the dynamics around the\nreference trajectory: this method is amenable to distributed communication\nimplementation where only one broadcast transmission is needed per sample.\nAlso, we provide closed-form expressions for instability and stability regions\nand convergence rate in terms of proportional gain $k$ and sampling period $T$.\nWe test the proposed control strategy via numerical simulations in the scenario\nof cooperative aerial manipulation of a cable-suspended load using a realistic\nsimulator (Fly-Crane). Finally, we compare our proposed controller with\ncentralized approaches that adapt the feedback gain online through smart\nheuristics, and show that it achieves comparable performance.\ndoi10.1016/j.automatica.2023.110941arxiv_id2112.00165pdf_urlshttps://arxiv.org/pdf/2112.00165githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, vol. 151, p. 110941, 2023categoriescs.RO cs.MA cs.SY eess.SY |
| titleNon-Sturmian sequences of matrices providing the maximum growth rate of\n matrix productsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsVictor Kozyakinabstract In the theory of linear switching systems with discrete time, as in other\nareas of mathematics, the problem of studying the growth rate of the norms of\nall possible matrix products $A_{\\sigma_{n}}\\cdots A_{\\sigma_{0}}$ with factors\nfrom a set of matrices $\\mathscr{A}$ arises. So far, only for a relatively\nsmall number of classes of matrices $\\mathscr{A}$ has it been possible to\naccurately describe the sequences of matrices that guarantee the maximum rate\nof increase of the corresponding norms. Moreover, in almost all cases studied\ntheoretically, the index sequences $\\{\\sigma_{n}\\}$ of matrices maximizing the\nnorms of the corresponding matrix products have been shown to be periodic or\nso-called Sturmian, which entails a whole set of \ properties of the\nsequences $\\{A_{\\sigma_{n}}\\}$, in particular the existence of a limiting\nfrequency of occurrence of each matrix factor $A_{i}\\in\\mathscr{A}$ in them. In\nthe paper it is shown that this is not always the case: a class of matrices is\ndefined consisting of two $2\\times 2$ matrices, similar to rotations in the\nplane, in which the sequence $\\{A_{\\sigma_{n}}\\}$ maximizing the growth rate of\nthe norms $\\|A_{\\sigma_{n}}\\cdots A_{\\sigma_{0}}\\|$ is not Sturmian. All\nconsiderations are based on numerical modeling and cannot be considered\nmathematically rigorous in this part; rather, they should be interpreted as a\nset of questions for further comprehensive theoretical analysis.\ndoi10.1016/j.automatica.2022.110574arxiv_id2112.00391pdf_urlshttps://arxiv.org/pdf/2112.00391githubhttps://github.com/kozyakin/barnormis_oasourcesopen-arxivtypejournaljournal_refAutomatica (2022) 110574categoriesmath.RA cs.NA math.NA |
| titleParallelized Robust Distributed Model Predictive Control in the Presence\n of Coupled State Constraintsyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsAdrian WiltzFei ChenDimos V. Dimarogonasabstract In this paper, we present a robust distributed model predictive control\n(DMPC) scheme for dynamically decoupled nonlinear systems which are subject to\nstate constraints, coupled state constraints and input constraints. In the\nproposed control scheme, all subsystems solve their local optimization problem\nin parallel and neighbor-to-neighbor communication suffices. The approach\nrelies on consistency constraints which define a neighborhood around each\nsubsystem's reference trajectory where the state of the subsystem is guaranteed\nto stay in. Reference trajectories and consistency constraints are known to\nneighboring subsystems. Contrary to other relevant approaches, the reference\ntrajectories are improved consecutively. The presented approach allows the\nformulation of convex optimization problems for systems with linear dynamics\neven in the presence of non-convex state constraints. Additionally, we employ\ntubes in order to ensure the controller's robustness against bounded\nuncertainties. In the end, we briefly comment on an iterative extension of the\nDMPC scheme. The effectiveness of the proposed DMPC scheme and its iterative\nextension are demonstrated with simulations.\ndoi10.1016/j.automatica.2024.111952arxiv_id2112.05965pdf_urlshttps://arxiv.org/pdf/2112.05965githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica (2025), Volume 171categorieseess.SY cs.SY |
| titleOn the Ergodic Control of Ensembles in the Presence of Non-linear\n Filtersyearvenue_abbrAutomaticavenue_nameAutomaticaauthorsVyacheslav Kungurtsev and Jakub Marecek and Ramen Ghosh and Robert N.\n Shortenabstract In many sharing-economy applications, as well as in conventional economy\napplications, one wishes to regulate the behaviour of an ensemble of agents\nwith guarantees on both the regulation of the ensemble in aggregate and the\nrevenue or quality of service associated with each agent. Previous work\n[Automatica, Volume 108, 108483, arXiv:1807.03256 ] has developed guarantees of\nunique ergodicity when there are linear filters. Here, we extend the guarantees\nto systems including non-linear elements, such as non-linear filters.\ndoi10.1016/j.automatica.2023.110946arxiv_id2112.06767pdf_urlshttps://arxiv.org/pdf/2112.06767githubis_oasourcesopen-arxivtypejournaljournal_refAutomatica, Volume 152, June 2023, 110946categoriesmath.OC |
| |
