PyBullet Robot

This package provides:

1. [BulletRobot]: A general Python interface class for robot simulations using PyBullet. Python API class to control and monitor the robot in the simulation.
2. [PybulletIKInterface] class: A constraint-based inverse kinematics (IK) "solver" interface. Instead of solving IK analytically, it attaches PyBullet constraints that pull the requested end-effector/link frames towards their target poses, forward-simulates the physics until they settle, and reads back the resulting joint configuration as the IK solution. It supports tracking several frames at once (full 6-DOF pose targets and position-only targets), and floating-base robots (where the base pose is solved together with the joint angles).

Installation

From PyPI

pip install pybullet_robot

From source

pip install git+https://github.com/justagist/pyrcf

Development

Install in editable mode with the development extras (test + lint + build tooling):

git clone -b main https://github.com/justagist/pybullet_robot
cd pybullet_robot
pip install -e ".[dev]"

A dev container is provided for a reproducible environment in VS Code ("Reopen in Container").

Usage

See examples/ for runnable demos covering different robots and capabilities. Robot models are downloaded automatically on first run via robot_descriptions:

cd examples
python demo_task_space_control.py

Examples

Joint position control

Loading a robot, inspecting it, position control, and reading state. (demo_joint_position_control.py)

Task-space control

Example using a custom cartesian impedance / torque control for task-space control. (demo_task_space_control.py)

Headless IK + separate-sim control

Run the provided PybulletIKInterface headless as a pure solver and apply its solution to a robot controlled via position control. (demo_ik_headless_control.py)

Whole-body / multi-end-effector IK

Floating-base IK tracking the body pose and all four feet simultaneously. (demo_quadruped_ik.py)

Interactive inverse kinematics (KUKA iiwa14)

Interactive IK: drag GUI sliders to set an end-effector target and the constraint-based solver reaches it. (demo_ik_interface.py) Visually similar to the headless IK demo above, so it has no separate clip; run it to try it interactively.