Research And Design Of Mobile Robot 3D Mapping And Navigation System Based On RGB-D Images

Localization and navigation are two key technologies in the research of mobile robots.Traditional localization methods rely on sensors in the environment,this imposes many limitations on application scenarios.Visual Simultaneous Localization and Mapping(V-SLAM)only requires the camera to determine the robot's position in the physical space and reconstruct the 3D structure of the scene.At the same time,with the advent of the low-cost RGB-D camera Kinect,it can directly perform 3D measurements on the environment and provide a lot of convenience for robot localization and navigation.In this context,mobile robot 3D mapping and navigation system based on RGB-D images is studied in this dissertation.The main research work includes:1.The line features are added to the traditional SLAM method,and 3D mapping algorithm based on point and line features is studied.Pose estimation and local map optimization are performed by minimizing the reprojection error of point and line features.Bag of words based on the point and line features is constructed for loop detection,globally consistent environment map is obtained by eliminating accumulated errors through closed-loop correction,and point cloud map is converted into an octree map.This method can effectively improve the robustness and accuracy of the 3D mapping algorithm and meet the requirements of subsequent navigation planning.2.In order to overcome the shortcomings of the traditional A* algorithm,the A* algorithm is improved through bidirectional search and smooth path,the efficiency of global path planning can be effectively improved and the number of turning times can be reduced.Combined with the advantages of global path planning and local path planning,a hierarchical path planning method based on improved A* and DWA algorithms is proposed.This method can optimize the path at the global level,avoid falling into local minimum,and correct the path in real time for dynamic environment to achieve dynamic obstacle avoidance.3.Mapping experiments through public datasets verifies the accuracy of 3D mapping algorithm based on point and line features.The indoor mapping experiments are conducted in the lab and the corridor respectively,and the path planning experiments are conducted in the laboratory.The experimental results verify the validity and reliability of the 3D mapping and path planning system proposed in this dissertation,and can achieve the functions of mapping and path planning in different environments.