This framework simplifies the development of reproducible robotics environments for reinforcement learning research, targeting RL researchers and engineers. It streamlines environment creation by abstracting common boilerplate code, allowing users to focus on decision-making logic and domain randomization.

How It Works

Gym-ignition leverages the ScenarIO project for low-level Ignition Gazebo simulator interfacing. It introduces Task and Runtime abstractions to manage environment setup and execution. The framework includes randomizers for domain randomization of models, physics, and tasks, and provides dynamics algorithms compatible with both fixed-base and floating-based robots using robotology/idyntree.

Quick Start & Requirements

• Install ScenarIO first.
• pip install gym-ignition (preferably in a virtual environment).
• Requires Ignition Gazebo.

Highlighted Details

• Facilitates reproducible robotics simulations for RL.
• Includes domain randomization features.
• Supports fixed-base and floating-based robot dynamics.
• Provides canonical example environments.

Maintenance & Community

This project is no longer actively maintained, with development stalled. For details on revival efforts, see robotology/gym-ignition#430. Community discussions are hosted on GitHub Discussions.

Licensing & Compatibility

LGPL v2.1 or any later version. This license generally permits use in closed-source applications, but modifications to the library itself must be shared under the LGPL.

Limitations & Caveats

The project is explicitly stated as no longer actively maintained and development has stalled, indicating potential issues with future support or bug fixes.