Humanoid robot navigates safely across multiple floors with lidar and learning
Learning Safe Humanoid Navigation from Reduced Order Models
Robotics
Summary
Humanoid robots find it hard to move safely across complex places like stairs or buildings with multiple floors. The authors show that teaching a robot all at once is too hard, so they split the problem into two parts. First, they teach a simpler model to navigate using 3D laser sensors, then use that knowledge to guide the full robot’s movement. Adding a safety check helps the robot avoid unexpected obstacles without lowering its success in reaching its destination.
What this means in practice
- •For robotics engineers: Enable humanoid robots to navigate buildings with stairs and multiple floors safely and efficiently without pre-made maps by combining simplified dynamics and laser sensing.
- •For warehouse automation teams: Deploy legged robots that can move autonomously across different floor levels and complex indoor environments using learned navigation policies and safety filters.$Commercial implications: Supports selling multi-floor autonomous robots capable of complex indoor navigation to logistics and warehouse providers.
Authors
William D. Compton, Zachary Olkin, Ryan Bena, Aaron D. Ames
Abstract
Research in humanoid robotics has achieved rapid progress in locomotion, and recent results have pushed the boundary on autonomous navigation. We demonstrate that a standard single-stage RL navigation pipeline struggles to scale to multi-level and multi-story terrain, limited by the difficulty of complex humanoid terrain interactions such as stairs. To overcome this challenge, we decompose the navigation problem into two pieces. First, we train a policy operating on the reduced order dynamics but with full 3D LiDAR observations to navigate complex, multi-story terrain. We then utilize this navigation knowledge to kickstart a policy operating on the full-order humanoid dynamics, with a frozen locomotion policy in the loop. Additionally, we demonstrate that applying a Poisson safety filter to the navigation policy output recovers safety in the presence of out-of-distribution obstacles, without dropping navigation success rate. We demonstrate the resulting RoM-Nav policy on a Unitree G1, accomplishing mapless multi-floor navigation covering trials with over 10m of vertical displacement and over 100m of path length. Project page with videos https://wdc3iii.github.io/rom-nav/ .