Reasearch On Mars Multi-rovers Cooperatively Constructing Navigational Maps

As Mars is very similar to the environment in which the Earth is located,it is of great significance to explore Mars in search of extraterrestrial life and to study the evolut ion of the Earth's atmosphere and magnetic fie lds and so on.The use of Mars rover to detect Mars is the most effective means for human detection so far.However,due to the lack of navigational maps on the surface of Mars,the current Mars rover 's travel route main ly relies on the design of the Earth's personnel and can not achieve autonomous path planning,which greatly reduces the detection speed of the Mars rover.In addition,the images captured by the Mars rover are used only for the navigation of a single rover,which does not provide help when other Mars rovers pass though the same area,result ing in waste of information.Based on these issues,this paper proposes a Mars mult i-rovers collaborative construction method of navigational map.The main research content is as follows.First,a SLAM(Simultaneous Localization and Mapping)method ba sed on stereo vision is studied.Using this method,A Mars rover is autonomously positioned and the sparse-points map of the surface of Mars is constructed,which can be used for autonomous positioning of other rovers when passing through the same area..Then,this paper studys the overlapping region search algorithm based on the bag of word,including the method of building a dictionary using the images captured by the Mars rovers,calculating the similarity score s of pictures,suppress the lower similarity scores,and finally the similar images of consecutive frames are found.Afterwards,this paper studys the method of stitching the sparse-points map constructed by different Mars rovers and the pose and position estimation of these rovers using the similar images of consecutive frames,and the results are nonlinearly optimized.Finally,this paper studys the octree map construction method for follow-up Mars rovers navigation applications,including the method of construcing the dense points map on Mars surface,optimiz ing the point cloud,and using the dense points map to contruct the octree map which can represent the space occupancy situation.This article has carried on the simulation experiment to each chapter content,demonstrated that the method proposed in each par t is accurate and effective.