REASAN: Learning Reactive Safe Navigation for Legged Robots

TL;DR

REASAN is a modular, real-time, reactive navigation system for legged robots that uses a single LiDAR sensor and reinforcement learning modules to navigate complex dynamic environments safely.

Contribution

The paper introduces a novel modular framework combining RL policies and a transformer-based estimator for legged robot navigation, trained without heuristics or complex policy switching.

Findings

Achieves onboard, real-time navigation in complex environments.

Demonstrates improved robustness over existing methods.

Validates modular design through comprehensive ablations.

Abstract

We present a novel modularized end-to-end framework for legged reactive navigation in complex dynamic environments using a single light detection and ranging (LiDAR) sensor. The system comprises four simulation-trained modules: three reinforcement-learning (RL) policies for locomotion, safety shielding, and navigation, and a transformer-based exteroceptive estimator that processes raw point-cloud inputs. This modular decomposition of complex legged motor-control tasks enables lightweight neural networks with simple architectures, trained using standard RL practices with targeted reward shaping and curriculum design, without reliance on heuristics or sophisticated policy-switching mechanisms. We conduct comprehensive ablations to validate our design choices and demonstrate improved robustness compared to existing approaches in challenging navigation tasks. The resulting reactive safe navigation (REASAN) system achieves fully onboard and real-time reactive navigation across both single- and multi-robot settings in complex environments. We release our training and deployment code at https://github.com/ASIG-X/REASAN.