Curated list of environments compatible with OpenAI gym

• gym-unity: Make Unity's ml-agents compatible with the gym interface

• OpenAI roboschool: Free robotics environments, that complement the Mujoco ones
• pybullet_env: Examples environments shipped with pybullet. From the official documentation:

PyBullet versions of the OpenAI Gym environments such as ant, hopper, humanoid and walker. There are also environments that apply in simulation as well as on real robots, such as the Ghost Robotics Minitaur quadruped, the MIT racecar and the KUKA robot arm grasping environments.

• pybullet-gym: Reimplementation of openAI gym Mujoco and Roboschool envs using pybullet. Planned support for Deepmind dm_control suite
• marathon-envs: Continuous control environments (similar to OpenAI's mujoco envs) in Unity

• obstacle tower env: The environment coming with Unity's obstacle tower challenge

• gym-duckietown: Duckietown self-driving car simulator environments for OpenAI Gym
• minerl: Find a diamond for in Minecraft for the MineRL NeurIPS 2019 challenge