Research On Mars Terrain Perception Based On 3D Point Cloud Information

At present,the Mars rover that has successfully landed on Mars and operated is the basic structure of six-wheel independent drive,main and vice rocker arms.The six-wheel independent drive not only provides a redundant backup function,but also brings about the problem of overdrive.It is desired to research on coordinating the speed of the six wheels.Based on this background,this paper proposes a method of using lidar to obtain three-dimensional point cloud information around the rover to build a global point cloud map and real-time perception of the terrain under the current wheel of the rover,so as to achieve wheels speed coordination.The main research content of this paper is to establish a global point cloud map based on the three-dimensional point cloud information and achieve real-time perception of the terrain under the wheel.This article first introduces the characteristics of mechanical lidar and it’s output three-dimensional point cloud,and points out the ordered information contained in the three-dimensional point cloud information.According to the characteristics of Martian terrain,multi-plane features are proposed,and based on this,a three-dimensional point cloud filtering and feature extraction algorithm is designed to quickly and effectively extract multi-plane features from a single frame of original point cloud.Then,the multi-frame point cloud registration is performed,and a comprehensive point cloud map construction algorithm combining a low-precision registration algorithm based on feature points and a high-precision improved GICP(Generalized Iterative Closest Point)registration algorithm is designed,which achieve a high-precision laser odometry and build a high-precision global point cloud map.The pose output by the registration algorithm based on feature points is used as the initial value of the GICP registration algorithm,which effectively improves the speed,accuracy and reliability of the GICP algorithm.The GICP algorithm is optimized by using the point cloud ordered information,which further improves the registration speed and ensures that the entire point cloud map construction algorithm can run in real time.Finally,the terrain perception and experimental verification under the Mars wheel were carried out,and the Mars environment simulation platform based on Gazebo and ROS and the experimental simulation site of the Mars rover prototype were built.The accuracy of the point cloud map construction and the feasibility of terrain perception under-wheel were verified from both simulation and experiment.