Imitation_Learning

Imitation Learning Algorithms (and IRL) based on RL library mushroom_rl.

• IL Implemented Algorithms: GAIL, VAIL

  • PPO or TRPO policy update versions available;
  • Option for easy states/actions selection for descriminator;

• IRL Implemented Algorithms: SCIRL

  • Examples for both MountainCar and Pendulum-v0 environments;
  • Feature Expectation estimator with Monte Carlo rollouts;
  • Estimation with Least-Squares Temporal difference is not debbuged (TODO);

• There are multiple examples available;

• Expert trajectories for some evironments;

• Algorithms are still being benchmarked in multiple environments;