Summary: Reinforcement learning benchmarks

这是一篇强化学习benchmark的总结，按照任务对其进行分类，并不全面地列举了在benchmark测试过的文章及开源代码。

2D navigation

gym-minigrid

• source code: https://github.com/maximecb/gym-minigrid
• scene screenshot:

• use cases:
  • Guiding Policies with Language via Meta-Learning

gym-maze

• source code: https://github.com/MattChanTK/gym-maze
• scene screenshot:

• use cases:

  • 3D navigation

    gym-miniworld

• source code: https://github.com/maximecb/gym-miniworld
• scene screenshot:

• use cases:
  • Decoupling Exploration and Exploitation for Meta-Reinforcement Learning without Sacrifices

Reference:

https://github.com/clvrai/awesome-rl-envs#navigation