About

This Repository contains implementation of ICM( Intrinsic Curiosity reward ) module by plugging it into three different algorithms i.e DQN , A3C and PPO. This whole code is based on the ICML paper introduced by Deepak Pathak,Pulkit Agrawal,Alexei A. Efros,Trevor Darrell in the paper Curiosity-driven Exploration by Self-supervised Prediction. ( For more information visit https://pathak22.github.io/noreward-rl/ ).

Organization

The repo is divided into 3 parts corresponding to each algorithm implemented. PPO and A3C is a group of python files in a folder while DQN is a jupyer notebook. For running just clone the repo and execute the code normally.

Demo for PPO + ICM:

1. Google Drive: https://drive.google.com/drive/folders/193MKIdXARZqCkVdSybdPXOQoEbfouXQB?usp=sharing

References:

1. A3C + ICM: https://youtu.be/O5Z-3q-J18I?si=bg4ytpGm63BtvltE
2. PPO + ICM: https://github.com/adik993/ppo-pytorch
3. https://pathak22.github.io/noreward-rl/