XLeRobot MuJoCo Simulation Controller

A Python-Mujoco keyboard controller for the XLeRobot

Overview

This controller provides keyboard controller for mobile chassis movement for omni-wheels and dual arm joint motion

Installation

1. Clone or navigate to the XLeRobot repository:

   cd /path/to/XLeRobot/simulation/mujoco
   

2. Install dependencies:

   python3 -m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt
   

   Key dependencies include:

   • mujoco==3.3.0 - Physics simulation engine
   • mujoco-python-viewer==0.1.4 - 3D visualization
   • glfw - Keyboard input handling

Usage

Basic Execution

python xlerobot_mujoco.py

The simulation will start with:

• A 3D viewer window showing the robot
• Real-time control interface
• On-screen control instructions

Control Scheme

🚗 Chassis Movement (Omnidirectional)

Action	Key	Description
Forward	Home	Move in +X direction
Backward	End	Move in -X direction
Left	Delete	Move in +Y direction
Right	Page Down	Move in -Y direction
Rotate CCW	Insert	Rotate counter-clockwise (+Z)
Rotate CW	Page Up	Rotate clockwise (-Z)

🦾 Left Arm Control

Joint	Positive	Negative	Description
Joint 1	Q	A	Shoulder rotation
Joint 2	W	S	Shoulder elevation
Joint 3	E	D	Elbow rotation

🦾 Right Arm Control

Joint	Positive	Negative	Description
Joint 1	U	J	Shoulder rotation
Joint 2	I	K	Shoulder elevation
Joint 3	O	L	Elbow rotation

Control Behavior

• Chassis Movement: Incremental velocity control with smooth acceleration/deceleration
• Arm Joints: Direct position control with small incremental steps (0.005 rad)
• Real-time Feedback: Live display of commanded vs actual values
• Smooth Dynamics: Automatic velocity decay when keys are released

Architecture

Class Structure

class XLeRobotController:
    def __init__(mjcf_path)      # Initialize simulation and viewer
    def update_feedback()        # Read sensor data from simulation
    def update_keyboards()       # Process keyboard input via GLFW
    def update_reference()       # Compute control commands
    def update_control()         # Apply commands to actuators
    def render_ui()              # Update 3D visualization and overlays
    def run()                    # Main control loop
    def cleanup()                # Graceful shutdown

Control Flow

1. Initialization: Load MJCF model, setup viewer and camera
2. Main Loop:
   • Read joint positions and velocities
   • Process keyboard inputs
   • Update control references
   • Apply controls to simulation
   • Step physics simulation
   • Render visualization (at 60 Hz)
3. Cleanup: Close viewer on exit