search

Breakend / gym-extensions

This repo is intended as an extension for OpenAI Gym for auxiliary tasks (multitask learning, transfer learning, inverse reinforcement learning, etc.)
https://breakend.github.io/gym-extensions/
Other
213 stars 41 forks source link

falcon landing demo #3

Closed vBarbaros closed 7 years ago

vBarbaros commented 7 years ago

added some docs files

Breakend commented 7 years ago

Also, this seems to have conflicts with the current master branch, can you look into resolving those after you've removed the commented out code and stuff?

vBarbaros commented 7 years ago

@Breakend Done, Thanks! (...and let me know if you'd like me to change something/reformulate the way I described the falcon-rocket env in the docs )

Breakend commented 7 years ago

Also, side note, might be interesting to add a fuel cost (like a time penalty?) to the reward. Not sure if this is there already.

Breakend commented 7 years ago

Btw, the suggestions about how to change the env aren't really for this pull request, just suggestions. I would say the only change I would suggest before merging in this request is just add a link to your page and maybe some wording in the docs text

vBarbaros commented 7 years ago

check lines 355-356, there is the penalty deducted from side_power and main_power consumed; one thing that is not added yet, though, is interrupting the episode Done=True if the rocket gets out of the x-axis bounds, will do that asap

Breakend commented 7 years ago

Looks awesome now! Very SpaceX. Could you also update the pic? at gym-extensions/raw/master/assets/Falcon.png and then i'll merge it in?

vBarbaros commented 7 years ago

oh! right

vBarbaros commented 7 years ago

done! Thanks for all the feedback, Peter!