Point cloud registration algorithm for autonomous landing based on color and intensity information

Abstract

The autonomous landing of unmanned helicopter is one of the necessary technical means to complete modern and complex tasks. Aiming at the problems such as poor real-time performance and little content in the process of acquiring terrain information, we proposed a multi-information point cloud registration algorithm. This algorithm integrates the color information and echo intensity information of the point cloud into the traditional registration algorithm and solves the problems of poor registration accuracy and convergence speed when the traditional algorithm deals with the point cloud. In order to further verify the proposed algorithm, the performance of different registration algorithms was evaluated and compared on the ford campus data set provided by the University of Michigan. The final results show that the proposed algorithm has the advantages of high precision and fast speed compared with the traditional algorithm.