Research On Adaptive Switch Method Of Lidar Point Cloud Frame Matching In Warehouse Environment

Simultaneous Localization and Mapping(SLAM)technology is mostly used to realize the autonomous positioning and navigation of Automated Guided Vehicle(AGV)in warehouse environment.In the structure of the Laser SLAM algorithm,frame matching is the core step,and its matching accuracy will directly affect the positioning accuracy of the SLAM algorithm.In this dissertation,in a warehouse environment with both shelf area and transfer area,an adaptive switching Iterative Closest Point(ICP)algorithm and Correlation Scan Match(CSM)algorithm based on Sigmoid function are studied and implemented.The main research contents are as follows:(1)According to the structural characteristics of the warehouse environment,it is divided into two areas: the shelf area and the transfer area.For these two areas,the front-end framework is constructed on the basis of the graph-optimized Laser SLAM algorithm,and the ICP algorithm and the CSM algorithm are used as the adjacent laser radar.Matching method of frame point cloud data,and introduces loop closure detection and back-end optimization(2)The characteristics of the warehouse environment are analyzed,and the number of feature points in the environment and the distance between the AGV and the surrounding objects are used as the characteristic parameters to characterize the shelf area and the transfer area.Aiming at the problem that the ICP algorithm and the CSM algorithm have different application areas in the warehouse environment,an adaptive switching technology based on the sigmoid function is studied.Using the Sigmoid function as the switching function,the number of feature points in the environment where the AGV is located and the distance value between the AGV and surrounding objects are used as the input of the switching function,and the output value of the switching function is used to realize the self-adaptation between the ICP algorithm and the CSM algorithm.This improves the matching accuracy of the SLAM algorithm in the warehouse environment.(3)Using lidar,IMU and odometry as sensors,NVIDIA TX2 as the main controller built the AGV experimental platform,and established the AGV coordinate system and kinematics model.In a single environment,it is verified that the ICP algorithm has the advantage of high matching accuracy in the narrow shelf area with many feature points,and the CSM algorithm has the advantage of high matching accuracy in the empty transit area with few feature points.Matching and localization experiments are carried out in complex warehouse environment to verify the effectiveness of the adaptive switching technology proposed in this dissertation.