티스토리 뷰

keep9oing

강화학습 환경들

HTS3 2021. 4. 15. 19:48

 

Atari, Mujoco, Gym이런거 말고 더 없을까 해서 알아봄. 마리오같은 흔한 환경은 제외함

재밌는 환경들이 많으니 감상만 하셔도 될듯합니다 ㅎㅎㅎ

1. D4RL

다양한 환경들

sites.google.com/view/d4rl/home

 

D4RL

A collection of benchmarks and datasets for offline reinforcement learning

sites.google.com

주방 환경도 있음 ㅋ

To Bear

2. CHALET: Cornell House Agent Learning Environment

집안일 관련 환경 제공.

방사이를 옮겨다니면서 집안 가구들과 상호작용을 액션으로 할 수 있음

github.com/lil-lab/chalet

 

lil-lab/chalet

Cornell House Agent Learning Environment. Contribute to lil-lab/chalet development by creating an account on GitHub.

github.com

이런거나
이런거

3. VizDoom

영상기반 둠 플레이

github.com/mwydmuch/ViZDoom

 

mwydmuch/ViZDoom

Doom-based AI Research Platform for Reinforcement Learning from Raw Visual Information. :godmode: - mwydmuch/ViZDoom

github.com

이런 식으로

4. TORCS

자동차 레이싱 게임

sourceforge.net/projects/torcs/

 

TORCS - The Open Racing Car Simulator

Download TORCS - The Open Racing Car Simulator for free. TORCS, The Open Racing Car Simulator is a highly portable multi platform car racing simulation. It is used as ordinary car racing game, as AI racing game and as research platform.

sourceforge.net

A3C논문에 나온 환경

 

5. Nethack

게임같은데 솔직히 뭔지는 잘 모르겠음 페북연구팀이 만듬

 

github.com/facebookresearch/nle

 

facebookresearch/nle

The NetHack Learning Environment. Contribute to facebookresearch/nle development by creating an account on GitHub.

github.com

봐도 뭐하는건지 잘 모르겠다

6. Starcraft2

복잡한 게임상황을 단순화시켜서 보여주기도 한다.

github.com/deepmind/pysc2

 

deepmind/pysc2

StarCraft II Learning Environment. Contribute to deepmind/pysc2 development by creating an account on GitHub.

github.com

게임화면
이렇게도 볼 수 있음

7. Teach my agent

Automatic curriculum learning을 위한 환경 

github.com/flowersteam/TeachMyAgent

 

flowersteam/TeachMyAgent

Contribute to flowersteam/TeachMyAgent development by creating an account on GitHub.

github.com

ㅋㅋㅋ

8. Griddly

Grid 기반 게임 관련 환경

griddly.readthedocs.io/en/latest/

 

Griddly Docs — Griddly 1.0.0 documentation

© Copyright 2020, Chris Bamford Revision 2934c86e.

griddly.readthedocs.io

9. Alchemy

메타 강화학습의 벤치마크용으로 나온 환경

github.com/deepmind/dm_alchemy

 

deepmind/dm_alchemy

DeepMind Alchemy task environment: a meta-reinforcement learning benchmark - deepmind/dm_alchemy

github.com

GIF인데 용량 초과라 스샷만 ㅠ

 

댓글
공지사항
최근에 올라온 글
최근에 달린 댓글
Total
Today
Yesterday
링크
«   2024/05   »
1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31