CrossLoco, a guided unsupervised reinforcement learning framework that simultaneously learns robot skills and their correspondence to human motions. The key innovation is to introduce a cycle-consistency-based reward term designed to maximize the mutual information between human motions and robot states.

[paper] [website]

Although the orginal version of CrossLoco is implemented based on LeggedGym, we now move our implementation to Orbit. Currently, this repo is still under construction but the core reward and algorithm is uploaded.

Install Orbit follow the instruction: https://isaac-orbit.github.io/orbit/source/setup/installation.html

Clone this repo under Orbit folder, then it should be fine.

The environment and training configuration are included in ./envs

The main algorithm and runner files are locates in ./runners