benelot / pybullet-gym

Open-source implementations of OpenAI Gym MuJoCo environments for use with the OpenAI Gym Reinforcement Learning Research Platform.
https://pybullet.org/
Other
814 stars 124 forks source link

How are the environments related with those in bullet3? #33

Closed diegoferigo closed 4 years ago

diegoferigo commented 4 years ago

@benelot In the upstream bullet3 repository I found https://github.com/bulletphysics/bullet3/pull/1296 and https://github.com/bulletphysics/bullet3/pull/1512. If I understood well, some of the work stored in this repo has been merged upstream. How does pybullet-gym differ from the upstream environments? Which ones should be preferred? Is there any plan about maintenance and improvements?

benelot commented 4 years ago

Hi! So there are some that are merged upstream as you correctly found out. In my case I am working on making them more compatible with the ones of the original OpenAI gym, but work progresses slowly towards that. As such, for now they can be considered nearly identical. There are some new envs coming up that are not upstream though. I plan to do maintenance of these envs here and people supporting me with this are highly welcome!

diegoferigo commented 4 years ago

Hi @benelot, thanks for shedding some light upon it, got it. I'll keep having a look to the development.