Environments

The Stable Gym package provides a variety of environments for training and testing (stable) reinforcement learning (RL) algorithms.

Biological environments

Gym environments that are based on Biological systems.

• Oscillator gymnasium environment
• Oscillator Complicated gymnasium environment

Classic control environments

Environments that are based on classical control problems or classical control environments found in the gymnasium library.

• Ex3EKF gymnasium environment
• CartPoleCost gymnasium environment
• CartPoleTrackingCost gymnasium environment

Mujoco environments

Environments that are based on the on Mujoco or Mujoco gymnasium environments.

• AntCost gymnasium environment
• HalfCheetahCost gymnasium environment
• HopperCost gymnasium environment
• HumanoidCost gymnasium environment
• SwimmerCost gymnasium environment
• Walker2dCost gymnasium environment

Robotics environment

• FetchReachCost gymnasium environment
• MinitaurBulletCost gymnasium environment
• QuadXHoverCost gymnasium environment
• QuadXTrackingCost gymnasium environment
• QuadXWaypointsCost gymnasium environment

Note

The ROS robotic environments from the Stable Gym package have been relocated to a distinct package known as Ros Gazebo Gym. When integrating this package with stable RL agents, such as those available in the Stable Learning Control package, it’s crucial to configure the positive_reward argument as True.