Simple package to run locomotion policies on a Kangaroo robot from PAL Robotics (1st version) simulated in MuJoCo for sim-to-sim validation.
Policy training was performed using Isaac Lab with the environment from isaaclab_kangaroo.
You can install the package in a virtual environment via pip install -e ..
The folder script contains a simple policy runner (run_policy.py) and three example policies (inside script/policies). In the example the desired velocity is selected randomly.
The MuJoCo model of the Kangaroo robot is inside the folder mujoco_kangaroo_sim2sim/kangaroo_mujoco.
@article{amadio2025learning,
title={Learning to Walk with Hybrid Serial-Parallel Linkages: a Case Study on the Kangaroo Robot},
author={Amadio, Fabio and Li, Hongbo and Uttini, Lorenzo and Ivaldi, Serena and Modugno, Valerio and Hoffman, Enrico Mingo},
year={2025}
}