PyTorch implementation of TD3-GAIL tested on the following environments:
TD3-GAIL is the recovered reward function based reinforcement learning that fusion of TD3 and GAIL. TD3-GAIL shows good learning performance compared with GAIL and DDPG-GAIL.
- To train the TD3-GAIL, run the 'train.py'
- To test the trained TD3-GAIL, run the 'test.py'
Python 3.6
PyTorch 1.5.0
NumPy 1.19.5
gym 0.15.4
Roboschool 1.0.34
Pillow 7.0.0
[1] [Official TD3 Code] (https://github.com/sfujim/TD3)
[2] [DGAIL] (https://github.com/nikhilbarhate99/Deterministic-GAIL-PyTorch)