Abstract

Abstract. Our contribution presents a new perspective in the mathematical description of a rotating multi-beam LiDAR sensor, in a sense that we make use of projective geometry along with the “homogeneous general equation of the second degree” to parametrize scan lines. We describe the scan geometry of a typical multi-beam rotating 3D LiDAR by representing scan lines as pojective conics that represent a projective figure (a cone) in an embedding plane. This approach enables the parameterization of each scan line using a generic conic section equation. Most modeling approachs model spinning LiDAR sensors in terms of individual points sampled by a laser beam. On the contrary, we propose a model that provides a high-level geometric interpretation both for the environment and the laser scans. Possible application scenarios include exterior and interior calibration of multiple rotating multi-beam sensors, scan distortion correction and localization in planar maps.

Highlights

  • Light Detection and Ranging (LiDAR) technology has been rapidly advancing due to an increasing demand for 3D mapping in numerous applications such a s autonomous driving and service robotics

  • Multi-beam rotating LiDAR sensors are popular for use in autonomous mobile mapping and robotics

  • Multi-beam rotating LiDAR sensors are constructed by positioning multiple range finders each positioned at a unique angle from the horizontal plane and all spinning about the z axis at the same rate

Read more

Summary

INTRODUCTION

Light Detection and Ranging (LiDAR) technology has been rapidly advancing due to an increasing demand for 3D mapping in numerous applications such a s autonomous driving and service robotics. Multi-beam rotating LiDAR sensors are popular for use in autonomous mobile mapping and robotics. This is primarily due to their compact form factor and their capability to provide three-dimensional scans typically under 0.1 seconds. Velodyne sensors are one of the most popular LiDAR sensors of this type. In order to use such sensors in different applications such as Multi-sensor calibration, Indoor SLAM and change detection, a unified mathematical model is deemed necessary. The empirical discussion is provided based on both a simulated data and real sensor data from the Velodyne VLP-16 LiDAR sensor

LiDAR Sensor Models
Geometric Preliminaries
Scan Mechanism
APPROACH
Stationary scans
Planar maps
Deterministic approach
Embedding a cone in a plane
Projectivity between conics
Rotating Multi-beam LiDAR Model
EXPERIMENT
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call