Quadruped Robot Learning and Navigation
Sim-to-real locomotion and long-range embodied navigation on Unitree GO2.
Overview
This work focuses on transferring learned control policies from simulation to a Unitree GO2 quadruped. Policies are trained and evaluated in Isaac Lab before deployment on the physical robot.
Contributions
The project includes staircase locomotion deployment, point-cloud processing, environment perception, motion-policy adaptation, and system-level debugging. It also integrates the LaViRA framework for long-distance embodied navigation.