Along Similar Lines: Local Obstacle Avoidance for Long-term Autonomous Path Following

TL;DR

This paper enhances a LiDAR-based VT&R3 system for long-term autonomous path following by improving obstacle detection and avoidance, using a new planning approach that generates smooth, terrain-aware paths in changing environments.

Contribution

It introduces a simplified change detection method for obstacle perception and a novel motion planner with a curvilinear space and edge-cost metric for better obstacle avoidance.

Findings

Reliable obstacle detection in changing environments

Smooth, terrain-aware path planning demonstrated

Effective long-term autonomous operation shown

Abstract

Visual Teach and Repeat 3 (VT&R3), a generalization of stereo VT&R, achieves long-term autonomous path-following using topometric mapping and localization from a single rich sensor stream. In this paper, we improve the capabilities of a LiDAR implementation of VT&R3 to reliably detect and avoid obstacles in changing environments. Our architecture simplifies the obstacle-perception problem to that of place-dependent change detection. We then extend the behaviour of generic sample-based motion planners to better suit the teach-and-repeat problem structure by introducing a new edge-cost metric paired with a curvilinear planning space. The resulting planner generates naturally smooth paths that avoid local obstacles while minimizing lateral path deviation to best exploit prior terrain knowledge. While we use the method with VT&R, it can be generalized to suit arbitrary path-following applications. Experimental results from online run-time analysis, unit testing, and qualitative experiments on a differential drive robot show the promise of the technique for reliable long-term autonomous operation in complex unstructured environments.