Enhance mujoco_server.py and README.md for improved teleoperation and environment management
Georg
9078db7
- Initialized last_teleop_command with default values to ensure consistent command structure.
- Updated switch_robot function to handle gripper state for UR5 robots, improving teleoperation control.
- Refactored the /env endpoint to return static environment information, including camera feeds and robot properties, enhancing API usability.
- Revised README.md to reflect changes in dependencies and clarify the usage of the new /env endpoint for environment information.
- Added detailed documentation on the new camera feed structure and updated examples for better clarity.
Enhance UR5e robot control with orientation tracking and UI updates
- Added support for end-effector orientation control in the UR5e robot, allowing for 6-DOF inverse kinematics.
- Implemented new WebSocket messages for setting target orientation and toggling orientation usage.
- Updated mujoco_server.py to broadcast end-effector orientation and target orientation in state messages.
- Enhanced the UI to include orientation sliders and display for end-effector orientation.
- Refined README.md to document new features, including orientation control and updated JSON message structures.
Update QuadrupedPyMPCController for enhanced strafe motion dynamics
- Increased strafe amplitude and added extra knee lift during strafe for improved movement.
- Refined strafe behavior by differentiating between leading and trailing legs, optimizing hip roll adjustments during both swing and stance phases.
- Enhanced comments for clarity on strafe mechanics and adjustments.
Enhance Spot robot control by integrating PyMPC and MPC Gait controllers
- Updated SpotEnv to support a new controller type parameter, allowing selection between 'mpc_gait', 'pympc', and 'trot'.
- Implemented Quadruped-PyMPC and MPC Gait controllers for improved gait management and stability.
- Added Spot robot configuration file for gym-quadruped integration.
- Introduced new controller files for MPC Gait and Quadruped-PyMPC, enhancing the robot's movement capabilities.
Refactor Spot and G1 robot model paths and add new assets
- Updated the path for loading the RL policy in G1 to use local assets.
- Refactored SpotEnv to load the model from local assets instead of MuJoCo Menagerie.
- Added new model files and assets for the Boston Dynamics Spot robot, including various OBJ files for its components.
- Created a changelog and license for the Spot model assets.
Add Docker support with Dockerfile, docker-compose.yml, and .dockerignore
- Introduced Dockerfile for building the application with necessary system dependencies for MuJoCo and OpenCV.
- Added docker-compose.yml to define the service configuration for the Nova Sim application.
- Created .dockerignore to exclude unnecessary files from the Docker build context.
- Updated SpotEnv and TrotGaitController comments for clarity on standing pose definitions.
Enhance SpotEnv initialization by adding framebuffer size overrides for higher resolution rendering.
Features:
- Unitree G1 humanoid (29 DOF) with RL walking policy
- Boston Dynamics Spot quadruped (12 DOF) with trot gait controller
- Web-based video streaming interface
- WebSocket state/command communication
- Interactive camera controls
- Robot switching without restart
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
