Anisotropic Models of Lidar Measurements in Voxel-based Scan Matching

Jason H. Rife

Peer Reviewed

Abstract:	This paper introduces anisotropic models for measurement error that describe voxel-based lidar scan matching. Scan-matching algorithms operate on a pair of lidar range-images and estimate a rigid transformation that aligns the static portions of both images. Most scan-matching algorithms consider each lidar sample as a three-dimensional point that provides registration information uniformly in all dimensions; however, in truth, each sample is simply a range measurement in one dimension. Scan registration in the lateral direction (normal to the lidar raypaths), requires correlation across multiple samples. This paper acknowledges the distinction between the ranging and lateral directions in lidar scan-matching and develops distinct measurement-error models for each. These anisotropic models are validated using data from an urban roadway. The validation process indicates a precise model for the lateral direction and a less precise (but bounding) model for the ranging direction.
Published in:	Proceedings of the ION 2026 Pacific PNT Meeting April 13 - 16, 2026 Hilton Waikiki Beach Honolulu, Hawaii
Pages:	201 - 215
Cite this article:	Rife, Jason H., "Anisotropic Models of Lidar Measurements in Voxel-based Scan Matching," Proceedings of the ION 2026 Pacific PNT Meeting, Honolulu, Hawaii, April 2026, pp. 201-215. https://doi.org/10.33012/2026.20626
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In