Research On Scene Reconstruction Of Lunar Surface And Path Planning Method Of Lunar Rover

For all the countries superior in space power,prospecting the lunar surface with rovers has become a hot spot in lunar exploration.As a special mobile robot for lunar surface,the lunar rover has excellent sports performance and carries a number of payloads.It can perform various scientific experiments such as lunar rock detection and lunar soil sampling.Perceiving the unknown lunar environment is an indispensable step in performing a probing task for lunar rover.On the basis of the known lunar environment,selecting the probe points and planning the motion path of the lunar rover is the key to the success of the probing mission.In this paper,based on the cooperation project with the China Academy of Space Technology named "Intelligent Strategy Analysis Software for Sampling and Packaging of Lunar Soil",some key technologies including the scene reconstruction technology,the probe point selection strategy and the path planning method of the lunar rover are studied.The main research work is as follows:Firstly,the 3D scene reconstruction method of lunar surface based on stereo vision is studied.The stereoscopic version model is established by the triangulation method of binocular camera.In order to obtain the disparity map,the image preprocessing method and the stereo matching method are analyzed.It is determined that the image is preprocessed by histogram equalization and bilateral filtering,and the region-based stereo matching method is adopted.Then,a dense disparity map is obtained through the MRF framework.After obtaining the point cloud of the lunar surface,in order to reconstruct the three-dimensional model of the lunar surface,a rapid modeling method based on KD tree for Delaunay triangulation is proposed,which effectively improves the reconstruction efficiency of the 3D scene of the lunar surface.Secondly,for the lunar rock and soil exploration missions,the probe point selection strategy considering multi-optimization performance indicators is established.Based on the 3D model of the lunar surface,by analyzing the structural characteristics of the lunar rover,a feasible cost function is established to satisfy the traveling conditions of the lunar rover.Based on the cost function,the safe motion area accessible by the lunar rover is obtained by the breadth first search algorithm.Aiming at the lunar rock and soil exploration missions of the lunar rover,by analyzing the influence of the flatness and illumination conditions of the lunar surface on the detection instrument,a multi-optimized performance index adapted to the probe point selection is constructed.In order to obtain the optimal probe point,based on the multi-optimization performance index of the probe point,a selection strategy of probe points based on particle swarm optimization is proposed.Then,in order to make the lunar rover reach the location of the detection point safely and efficiently,the path planning method of the lunar rover is studied in this paper.By integrating the slope path cost and roughness cost,a path cost function based on the 3D model of the lunar surface is established,which can solve the characterization of the path cost during the movement of the lunar rover.Based on the traditional PRM algorithm,an improved method based on virtual force field is proposed,which effectively solves the problem of narrow passages existing in traditional PRM.The improved PRM improves the efficiency of constructing a roadmap with a high connectivity and its applicability.A path planning algorithm for a lunar patrol was established based on this improved PRM algorithm.Finally,the experiments of the scene reconstruction technology,the probe point selection strategy and the path planning method are discussed.A semi-physical experimental platform was built,and the experimental process was designed.The experiments are carried out,and the experimental results were compared and analyzed to verify the correctness and effectiveness of the proposed algorithm.