Target Localization and Following Based on LiDAR and Ultra-Wideband Ranging with Consideration of Target Visibility

Abstract

To perform target-following tasks in unknown environments, a robot must identify the target’s position and plan an efficient path to reach it. Traditional LiDAR-based localization systems face challenges in distinguishing the target from objects with similar appearances. Meanwhile, existing target-following approaches often neglect target visibility during path planning, leading to target occlusion by obstacles and ultimately resulting in following failure. In this paper, we propose a sequence matching method for target-localization using LiDAR and Ultra-Wideband (UWB) ranging. We determine the position of the target by analyzing the similarities between UWB ranging sequence and LiDAR cluster trajectories. To achieve visibility-aware target-following, we incorporate a visibility objective function into the Dynamic Window Approach (DWA) to generate a following path that minimizes the risk of target loss. This function evaluates the target loss risk based on the positional relationships between the robot, the target, and the nearest obstacle to the target. Extensive experiments were conducted using both human and robot as targets. The results show that our approach achieves higher completion rates when compared to the target-following using traditional DWA.