search

mit-acl / rl_collision_avoidance

Training code for GA3C-CADRL algorithm (collision avoidance with deep RL)
119 stars 30 forks source link

Question about how to make Learning cadrl! #19

Closed kmzdaniel closed 1 year ago

kmzdaniel commented 1 year ago

Hello, Mr. Everett. I added an Extended on your last ga3c code and tested it. and I need extra credit for my article. for that, I will be testing this Extended on your 'cadrl' model. but I can't find a policy like the 'LearningPolicyGA3C.py', Is there any way to make it?

mfe7 commented 1 year ago

Hi - I'm not sure I follow what you are asking.

kmzdaniel commented 1 year ago

Hello again - I need learning code of 'CADRL', like 'LearningPolicyGA3C.py'.

mfe7 commented 1 year ago

There is a pre-trained CADRL policy in the gym environment, same folder as the policy linked above.

If you are trying to train a new policy using the CADRL algorithm, we never put together a clean release of that training code. All the training code should be buried somewhere deep in the gym environment as well, but I can't speak to how easy it would be to run that (it's quite old).

kmzdaniel commented 1 year ago

Many thanks for your adequate explanation.