Overbound Models for Lidar Range Differences

Jason H. Rife

Peer Reviewed

Abstract:	This paper analyzes field data to quantify, for the first time, probability distributions of differences of lidar range-measurements taken along the same raypath over time. Range differences are important in quantifying the input measurement errors to lidar scan-matching algorithms. Scan-matching algorithms seek to align static features observed in two lidar images, in order to infer motion of the lidar unit between the two images. In essence, the lidar scan-matching algorithm estimates pose (translation and rotation) by comparing each range measurement in one image to a range measurement in the second image, with the goal of driving the difference to zero. In this sense, the noise in differencing two equivalent range measurements defines the input measurement error, which in turn determines the output error for pose estimation. This paper seeks to quantify this measurement error by analyzing data from four urban road intersections in San Francisco, USA. In all, over one million raypaths are analyzed to construct range-difference distributions. Most distributions were smooth, dominated by random electronic noise, but approximately 15% were multi-modal, with salient jumps between sample clusters. Nonparameteric models of the distribution are described, as well as conservative parametric models, in the form of a paired Gaussian overbounds.
Published in:	Proceedings of the 2026 International Technical Meeting of The Institute of Navigation January 26 - 29, 2026 Hyatt Regency Orange County Anaheim, California
Pages:	880 - 894
Cite this article:	Rife, Jason H., "Overbound Models for Lidar Range Differences," Proceedings of the 2026 International Technical Meeting of The Institute of Navigation, Anaheim, California, January 2026, pp. 880-894. https://doi.org/10.33012/2026.20516
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In