Computational Sensorimotor Learning HW1
Gabriel Margolis
Dependencies
The code will run in any python3 environment with airobot, gym, and tensorflow installed. Simply clone this repository onto your system and run the main training files.
Problem 1
The code for learning a policy for the reacher environment with PPO is in: learn_reacher_ppo.py
Problem 2
The code for learning a policy for the reacher_wall environment with PPO is in: learn_reacher_wall_ppo.py
Problem 3
The code for learning a policy for the pusher enviornment with PPO is in: learn_pusher_ppo.py
Acknowledgements
The gym environment templates used for training were provided by the 6.884 staff. Code for PPO training is adapted from OpenAI's stable baselines library.