Research On The Method Of Mobile Robot Map Construction Based On Multi Sensors

The map construction of mobile robots is the premise of mobile robots to realize autonomous navigation in the true sense,and it is also one of the most challenging problems in the research of key technologies of mobile robots.The method of constructing an environmental map using laser radar can basically satisfy a wide range of obstacles and single scenes.However,the scanning laser radar is a line,it can only obtain environmental information of a certain two-dimensional plane,environmental information will not be able to obtain three-dimensional space height,cause the mobile robot to build map is incomplete,inaccurate and lack important environment obstacle information,appear in the navigation of mobile robot is hit by mistake.In this thesis,a new method for constructing environment map based on multi-sensor fusion of laser radar and depth camera is proposed for the above problems.The main research contents are as follows:Firstly,the mobile robot coordinate system and kinematics model are established.The laser radar model and depth camera model are studied.The internal and external parameters of depth camera are calibrated.The common maps and models of mobile robots are introduced.Secondly,theoretical research and simulation analysis of SLAM map construction algorithm.The mathematical model of SLAM is established,and the principle and system structure of SLAM are introduced,which leads to the key problems of SLAM.Then,from the theoretical analysis,the Gmapping-SLAM algorithm,Hector-SLAM algorithm and Cartographer-SLAM algorithm are deeply studied,and three common construction graph algorithms were simulated in the ROS system.The simulation results show that the Gmapping-SLAM algorithm has good comprehensive performance.As the solution of the mobile robot SLAM in this thesis,it lays a foundation for the construction of environmental map based on multi-sensor fusion.Thirdly,multi-sensor fusion scheme research and design.Based on the complementary advantages of the two sensors of laser radar and depth camera,a new method for map building by multi-sensor fusion are proposed.This thesis studies the principle of transforming the depth map into a pseudo-laser in the multi-sensor fusion process,elaborates the new method of projecting the three-dimensional height information of the environment onto the two-dimensional plane information.The noise points and cavities in the depth map are processed by bilateral filtering.For the laser data transformed by laser radar and depth camera,Gmapping-slam algorithm was used to build the local raster map,and Bayesian method combined with map fusion rules is used to fuse local raster map.Finally,the laboratory mobile robot research and development platform is used to construct environmental maps in the real environment by using only laser radar and the new method of multi-sensor fusion proposed in this thesis.Experimental results show that the map constructed by the new method of multi-sensor fusion is more complete and accurate,which proves the feasibility of the new method.