Simultaneous Localization And Mapping Of AGV Based On LIDAR

Navigation technology is one of the core technology of AGV,and Simultaneous Localization and Mapping of AGV is the key to solve the navigation technology.FastSLAM is the most widely used algorithm in the current filed of Simultaneous Localization and Mapping.However,particle degradation and other issues still exist in the FastSLAM algorithm.It's significant to solve these problems of SLAM,because the research of technical theory and method about Simultaneous Localization and Mapping technology of mobile robot can drive the development of mobile robot system in construction,transport,industry,service industry and military and other fields.We conduct a study about the indoor environment map building based on laser radar sensor firstly,and complete the creation of a grid map by getting observation data,the ICP scan matching,data fusion.The basic steps of particle filter are studied and the particle filter algorithm is simulated.At the same time,the positioning function of the mobile robot based on particle filter is Carried out.Then,based on RBPF,we put forward and realized an improved adaptive algorithm on the basis of FastSLAM.Improved particle filter algorithm regularization in the course of resampling,and sampling from continuous distribution to maintain the diversity of particles.At the same time,using UKF algorithm instead of EKF algorithm to estimate the sign,UKF algorithm no need to linearized the nonlinear model,so it reduced the amount of calculation,through this method can improve the real-time and precision of the map building.Finally,Matlab simulation experiment proves that the improved resampling method increases the positioning accuracy compared with the traditional FastSLAM algorithm.Then,Gazebo is adopted to simulate the real environment and robot operation on ROS platform.We build two real-time environment map based on the traditional FastSLAM algorithm and the improved algorithm displayed on the Rviz shows that the improved algorithm build map with higher precision and better real-time performance.