How ir-lab/irl_control works
Project Type and Purpose
• This is a Python-based simulation and control *framework* for bimanual robot manipulation (dual-arm UR5) built on MuJoCo.
• It provides:
• Core APIs for robot abstraction, control algorithms (OSC, admittance), and simulation management.
• Pluggable input-device drivers (PS Move, Space Mouse) for teleoperation.
• YAML/XML-driven configuration for robot kinematics, PID gains, action sequences, and scene descriptions.
• Example scripts as user entry points, and a mesh/scene repository for 3D objects.
File Structure
Top-level:
• setup.py, requirements.in, README.md → packaging and dependencies
• mujoco_app.py, osc.py, robot.py, utils.py, osc.py, version.py → core library entry points
irl_control/ (core package):
• action_sequence_configs/*.yaml → high-level task definitions
• device.py → abstract Device class
• input_devices/ → concrete device drivers (ps_move.py, space_mouse.py)
• robot.py → Robot abstraction (kinematics, jacobians, state interfaces)
• mujoco_app.py → application loop: load scene, attach controllers, step sim, render
• osc.py → operational‐space control implementations
• utils.py → shared utilities (logging, data handling)
• meshes/, scenes/ → 3D asset repository and MuJoCo XML scene files
• robot_configs/*.yaml → per-robot and per-task PID/gain settings
• examples/ → demo scripts showing usage patterns
• tests/ → automated tests invoking run_tests.py
README for Insights
• Simulation backbone: MuJoCo (via mujoco-py)
• Extensible framework: add new input devices, new control algorithms, new mesh/scene assets
• Config-driven: YAML for gains, tasks; XML for scenes
• Demo-driven: examples illustrate how library pieces fit together
Be Detailed
• Versioning/test harness (version.py, tests/) in a peripheral “development & QA” section.