mjpl

MuJoCo motion planning library.

[!Note] This project is under active development. APIs may change, and features are still a work in progress.

Features:

• Joint-space planning via bi-directional RRT, with support for constraints
• Cartesian-space planning
• Interfaces for constraints, inverse kinematics, and trajectory generation

Limitations:

• This library is designed for manipulator models that are composed of hinge/slide joints, and will not work with models that have ball/free joints.

Installation

pip install mjpl

Development

For local development, clone the repository and install the developer dependencies:

pip install -e ".[dev]"

This project uses Ruff for linting and formatting. Unit tests are written via the unittest framework.

To run the linter, formatter, and tests locally:

# Check for linter errors
ruff check .
# Apply lint fixes
ruff check --fix .

# Check for format errors
ruff format --diff .
# Apply format fixes
ruff format .

# Run unit tests
python3 -m unittest -v

Pre-commit hooks are also available which run the linter, formatter, and unit tests. To trigger the hooks automatically on commit, install the pre-commit hooks:

pre-commit install

To trigger the hooks manually:

pre-commit run --all-files

To bypass installed pre-commit hooks on commit:

git commit --no-verify -m "your message"

Acknowledgements:

Thank you Sebastian Castro for the guidance and support that has been offered throughout the early stages of this project. If you find this library useful or interesting, consider checking out Sebastian's pyroboplan, which offers similar features via Pinocchio!