PERAL: Perception-Aware Motion Control for Passive LiDAR Excitation in Spherical Robots
Shenghai Yuan, Jason Wai Hao Yee, Weixiang Guo, Zhongyuan Liu, Thien-Minh Nguyen, Lihua Xie
AI summary
Problem
Horizontally mounted LiDARs on compact mobile robots suffer from poor near-ground coverage and blind spots, while existing active excitation methods add significant cost, weight, and power overhead.
Approach
The authors developed PERAL, a spherical robot that leverages the natural coupling between its internal differential-drive motion and shell attitude to passively rock the top-mounted LiDAR during locomotion, diversifying scan directions without extra hardware.
Key results
- 75.7% near-ground coverage outperforming fixed-horizontal baseline
- Reliable path-following across two shell sizes in manual and autonomous modes
- Successful traversal of 15° slope with perception-aware control
- Passive drive–shell coupling naturally diversifies LiDAR scan directions
Why it matters
Provides a low-cost, lightweight, and energy-efficient sensing enhancement strategy for compact mobile robots operating in feature-sparse or near-ground critical environments.
Abstract
Autonomous mobile robots increasingly rely on LiDAR–IMU odometry for navigation and mapping, yet hor- izontally mounted LiDARs often provide poor near-ground coverage, while alternative configurations risk geometric de- generacy. To address this, we present PERAL, a perception- aware spherical robot that enhances LiDAR viewpoint diversity through passive self-excitation, without dedicated scanning ac- tuators. By exploiting the coupling between internal differential- drive motion and shell attitude, PERAL naturally induces mild sensor rocking during locomotion, enriching vertical observa- tions. In controlled trials over comparable traveled distances, PERAL achieves (75.7%) coverage, substantially outperforming a fixed-horizontal differential-drive baseline (15.5%) and ap- proaching a quadruped platform with a bottom-mounted rotat- ing LiDAR (84.7%). Additional experiments demonstrate con- sistent path-following behavior across two shell sizes, reliable operation in both manual and autonomous modes, and success- ful traversal of a 15◦slope. These results suggest that passive self-excitation offers an effective and lightweight alternative to active LiDAR actuation for improving near-ground perceptual coverage on compact mobile robots. Design and code are available at https://github.com/snakehaihai/PERAL_ robot_design.