Research On Construction Method Of Logistics AGV High-precision Map Based On 2D Lidar

With the acceleration of China's industrialization process,the level of industrial automation continues to increase.If logistics,a labor-intensive industry,wants to reduce costs,it is even more important to improve automation.Logistics AGV(Automatic Guided Vehicle)is important transportation equipment in the indoor warehouse.Its autonomy and intelligence are at the core.From the perspective of industry development,positioning technology and mapping technology are the foundation of intelligence and autonomy.The experimental equipment used in this thesis is an autonomously built logistics AGV(named Robot),equipped with an odometer,lidar,and depth camera.A variety of SLAM(simultaneous localization and mapping)algorithms have been studied on the mobile robot.The first is to assemble the AGV,set the coordinate system,construct the relevant sensor models(compile function packages),select the environment map model,and adjust the speed.Secondly,the SLAM algorithm based on filter and graph optimization is introduced respectively,and the SLAM algorithm based on laser and the SLAM algorithm based on vision is briefly introduced.The SLAM algorithm based on particle filtering is represented by the Gmapping algorithm.The Gmapping algorithm incorporates odometer information.This type of algorithm has the advantages of high accuracy and small memory consumption when constructing small maps.The SLAM algorithm based on graph optimization is represented by the RGB-D SLAM algorithm.The front-end builds the map and the back-end performs loop detection.This is also the current mainstream visual SLAM algorithm framework.Large warehouse areas can often reach thousands of square meters,and the environment is complex.Traditional SLAM algorithms are no longer suitable for such environments.Based on the analysis of the advantages and disadvantages of various algorithms,the Google open-source Cartographer algorithm was used in 2016,and the algorithm was transplanted to the Robot robot used in this experiment.The idea of multi-sensor fusion was used to integrate the odometer Information,use visualization tools Rviz and simulation environment Gazebo to carry out simulation experiments,verify the effects of various algorithms applied to large environment maps and small environment maps,and focus on the introduction of Cartographer algorithm before and after fusion.Finally,experiments are performed on an autonomously constructed Robot mobile robot.The experimental results prove that the Cartographer algorithm after merging odometer information can construct a high-precision environment map,and the research is practical.