Research On AGV Mapping And Positioning Technology Based On Laser SLAM

As an automated transportation equipment,AGV is widely used in the fields of manufacturing,logistics,warehousing and other fields.It can replace traditional artificial transportation and has the advantages of high efficiency,low cost,and high safety.SLAM technology and global positioning technology in AGV equipped with lidar are research hotspots and difficulties in this field.This paper first analyzes the RBPF-SLAM algorithm based on particle filter,and proposes two improvements for the two problems of excessive operation cost caused by the fixed number of particles in the conventional algorithm and particle dissipation caused by too many sampling times.One is to improve the accuracy of the particles by incorporating the recent lidar data into the proposed distribution;second,the concept of adaptive resampling is introduced,and resampling is performed according to the high and low grouping of particle weights to alleviate the problem of particle dissipation.Secondly,this paper studies and analyzes the widely used positioning algorithm AMCL in mobile robots.Since the update of both Monte Carlo and adaptive Monte Carlo particles is based on the motion odometer,however,the wheel-type motion odometer alone will cause large errors over time.Therefore,aiming at the nonlinear system of AGV in this paper,the extended Kalman filter is used to fuse the IMU data and the motion odometer data as the motion model in the particle update process,which is beneficial to improve the accuracy of each particle,so as to improve the positioning accuracy of AGV.Finally,based on the ROS system,the simulation and real vehicle platform are built to compare the original RBPF-SLAM algorithm with the improved one,and the accuracy of the AMCL fused with gyroscope data is tested to verify the feasibility of the algorithm.The experimental data show that the improved RBPF-SLAM algorithm has higher accuracy and does better in the details of the map.The accuracy of the improved AMCL positioning algorithm is also improved compared with the original algorithm using a wheeled motion odometer.