Improving ICP Precision for Spacecraft Pose Estimation

ICP (iterative closest point) algorithm has been extensively applied for LIDAR (light detection and ranging) based spacecraft pose estimation. However the precision of estimated poses still suffers from point mismatches between LIDAR data and target model. In this paper, we find that by combining the laser direction of LIDAR, the number of mismatches could be reduced. Furthermore, robust estimation is introduced to ICP algorithm to alleviate influence of mismatches and outliers and its convergence is mathematically proved. Results on simulated data demonstrate that proposed strategies can significantly improve the precision of ICP.