Motion Estimation and Navigational Drift Correction with LiDAR Data

Rajaindra A. Balaebail and Kartik B. Ariyur

Abstract:	We develop a rigorous basis for motion estimation using LiDAR data, for the ultimate purpose of correcting drift in Inertial Navigation Systems (INS). This ensures that there is a predictable statistical relationship between uncertainty in algorithm input or sensor data and the uncertainty in algorithm output or motion estimates. Further, we can avoid the use of feature extraction algorithms and associated stochastic uncertainty. We have chosen the LiDAR as our relative navigation sensor for the following reasons: one-to-one correspondence between 3D point clouds and the physical world; accuracy and clean mathematical characterization. We consider two successive point clouds obtained from the LiDAR sensor. We compute the crosscorrelation between the corresponding point clouds using standard techniques. We posit the spatial volume in which the correlation function is non-zero as our error metric. Our results transform motion estimation and scene matching to area and volume minimization for 2D and 3D cases respectively. To extract relative motion, we first determine the position of the region where the output of the cross-correlation between the two point clouds is non-zero; the centroid of this region directly yields the translation. We determine rotation via minimizing the volume of the region where the correlation is non-zero. We accomplish this minimization through gradient descent, using our proof of the convexity of region volume with respect to the rotation angles. Using the properties of gradient descent and problem convexity, we arrive at the number of steps required to compute the actual translation and rotation given a specific desired accuracy and maximum rotation expected. The number of steps is proportional to the logarithm of the ratio of frame-to-frame rotation to the required angular accuracy. We handle clutter and dynamic objects using a voting mechanism whose properties we establish.
Published in:	Proceedings of the 2010 International Technical Meeting of The Institute of Navigation January 25 - 27, 2010 Catamaran Resort Hotel San Diego, CA
Pages:	167 - 175
Cite this article:	Balaebail, Rajaindra A., Ariyur, Kartik B., "Motion Estimation and Navigational Drift Correction with LiDAR Data," Proceedings of the 2010 International Technical Meeting of The Institute of Navigation, San Diego, CA, January 2010, pp. 167-175.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In