File size: 2,361 Bytes
0d603f4 195038d 0d603f4 195038d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 | ---
title: armteam
emoji: 🤖
colorFrom: gray
colorTo: indigo
sdk: static
pinned: false
---
# Advanced Robotic Manipulation
Models, datasets and evaluation artifacts for **contact-rich robot manipulation** — tactile
sensing, world-action models and real-robot evaluation on a UR3 with optical-tactile fingertips.
## HapticWAM — *Distilling Imagined Touch into a World–Action Model without Inference-Time Tactile Sensing*
A tactile world-action model whose teacher sees optical-tactile fingertips and generates a
structured **contact package** alongside its action chunk; the fingertip inputs are then distilled
away, so a **pad-free student** keeps imagining contact from vision, proprioception and wrist
force/torque alone.
📄 [Paper (arXiv:2609.23888)](https://arxiv.org/abs/2609.23888) ·
🌐 [Project page](https://advanced-robotic-manipulation.github.io/websites/hapticwam/) ·
💻 [Code](https://github.com/Advanced-Robotic-Manipulation/HapticWAM) ·
📦 [Collection: everything in one place](https://huggingface.co/collections/armteam/hapticwam-icra-2027-6ab234bdfe24c383b5f9fb57)
**Models** — [teacher](https://huggingface.co/armteam/hapticwam-teacher) ·
[student (deployed, pad-free)](https://huggingface.co/armteam/hapticwam-student) ·
[baselines](https://huggingface.co/armteam/hapticwam-baselines) ·
[ablations](https://huggingface.co/armteam/hapticwam-ablations)
**Datasets** — [teleop (packed training corpus)](https://huggingface.co/datasets/armteam/hapticwam-teleop-dataset) ·
[teleop raw](https://huggingface.co/datasets/armteam/hapticwam-teleop-raw) ·
[sim episodes](https://huggingface.co/datasets/armteam/hapticwam-sim-episodes) ·
[rig episodes](https://huggingface.co/datasets/armteam/hapticwam-rig-episodes) ·
[rollouts](https://huggingface.co/datasets/armteam/hapticwam-rollouts) ·
[results](https://huggingface.co/datasets/armteam/hapticwam-results)
## HapticVLA / Crab
Tactile-aware vision-language-action models and the bimanual mobile platform they run on.
💻 [Code](https://github.com/Advanced-Robotic-Manipulation/crab) ·
models: [left arm](https://huggingface.co/armteam/crab-smolvla-left-arm) ·
[right arm](https://huggingface.co/armteam/crab-smolvla-right-arm) ·
[RWFM](https://huggingface.co/armteam/crab-smolvla-rwfm) ·
[HapticVLA](https://huggingface.co/armteam/crab-smolvla-hapticsvla)
|