Vision-Based Reactive Planning and Control of Quadruped Robots in Unstructured Dynamics Environments(ICARM 2023)

Abstract

Empowering legged robots with agile maneuvers is a great challenge. While existing works have proposed diverse control-based and learning-based methods, it remains an open problem to endow robots with animal-like perception and athleticism. Towards this goal, we develop an End-to-End Legged Perceptive Parkour Skill Learning (LEEPS) framework to train quadruped robots to master parkour skills in complex environments. Specifically, a vision-based perception module with multi-layered scans provides robots with comprehensive and precise environment information. Based on the visual information, a position-based task formulation frees the robot from velocity constraints and steers the robot toward the target with novel rewards. The developed controller enables a low-cost quadruped robot to successfully overcome previously challenging and unprecedented obstacles. We evaluate LEEPS on various challenging tasks, which demonstrate its effectiveness, robustness, and generalizability. 

Method

Main Results

Reactive Planning

Reactive Control