Repolode
Submit
learnsyslab

learnsyslab/gym-pybullet-drones

PyBullet-based Gymnasium environments for training single and multi-agent reinforcement learning controllers for quadcopters.

View on GitHubWebsite
Stars
2,201
Forks
578
Open issues
111
Last commit
2 days ago

What is it?

gym-pybullet-drones provides quadcopter simulation environments built on PyBullet physics and the Gymnasium API. It lets researchers and students train reinforcement learning policies for single-drone and multi-agent control, such as hovering, formation flight, and downwash interactions between drones.

The project is a streamlined refactor of an earlier IROS 2021 codebase, updated for Gymnasium and Stable-Baselines3 2.0. It includes classic PID and velocity controllers, an MRAC adaptive controller, and example scripts for training PPO agents with learn.py and replaying them with play.py.

It also includes a Betaflight SITL example on Ubuntu for running multiple simulated flight controllers against the environment.

Key features

  • 01Gymnasium-compatible single-agent and multi-agent quadcopter environments
  • 02Built on PyBullet physics simulation
  • 03Includes PID, velocity control, and MRAC controller examples
  • 04Stable-Baselines3 PPO training and playback scripts
  • 05Downwash effect modeling between drones
  • 06Betaflight SITL integration on Ubuntu

Use cases

  • Training reinforcement learning controllers for quadcopter hover or trajectory tracking
  • Researching multi-drone coordination and formation flight
  • Testing control algorithms against Betaflight firmware in software-in-the-loop setups

Quick start

  1. 1

    Clone the repository

    git clone https://github.com/learnsyslab/gym-pybullet-drones.git
  2. 2

    Create and activate a conda environment with Python 3.12

    conda create -n drones python=3.12 && conda activate drones
  3. 3

    Install the package in editable mode

    pip3 install -e .
  4. 4

    Run a PID control example

    cd gym_pybullet_drones/examples/ && python3 pid.py

Things to know

  • Building pybullet on Python 3.11 and newer may need a compiler or a macOS-specific CFLAGS workaround.
  • The Betaflight SITL example works on Ubuntu only.

In collections

Open-source AI agent frameworks & appsReinforcement learning frameworks & training environmentsRobotics & embodied AI projects

Topics

#reinforcement-learning#quadcopter#drone#robotics#pybullet#gymnasium#multi-agent#simulation#betaflight#control

FAQ

What is gym-pybullet-drones?
PyBullet-based Gymnasium environments for training single and multi-agent reinforcement learning controllers for quadcopters. gym-pybullet-drones provides quadcopter simulation environments built on PyBullet physics and the Gymnasium API. It lets researchers and students train reinforcement learning policies for single-drone and multi-agent control, such as hovering, formation flight, and downwash interactions between drones.
Is gym-pybullet-drones free to use? What license does it have?
gym-pybullet-drones is open source under the MIT license. Check the license terms before using it in commercial products.
Is gym-pybullet-drones still maintained?
Actively maintained. The last commit to gym-pybullet-drones was on Nov 6, 2026.
What can I use gym-pybullet-drones for?
Training reinforcement learning controllers for quadcopter hover or trajectory tracking Researching multi-drone coordination and formation flight Testing control algorithms against Betaflight firmware in software-in-the-loop setups
How do I install gym-pybullet-drones?
Clone the repository: git clone https://github.com/learnsyslab/gym-pybullet-drones.git Create and activate a conda environment with Python 3.12: conda create -n drones python=3.12 && conda activate drones Install the package in editable mode: pip3 install -e . Run a PID control example: cd gym_pybullet_drones/examples/ && python3 pid.py
Where can I get gym-pybullet-drones?
The source code and installation instructions are on GitHub: https://github.com/learnsyslab/gym-pybullet-drones

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects