Research On Point Cloud Registration Algorithm Of Automatic Guided Vehicle

Research on Automated Guided Vehicle(AGV)can be directly mapped to the field of medical services.The purpose of this thesis is to build an AGV for medical services to cooperate with medical staff to deliver medicines and make rounds.In the special working environment of the hospital,due to the limitation of sterility and low noise,it is just right to use the AGV to cooperate with the medical staff to complete the related work.At this time,the AGV needs to have great Simultaneous Localization And Mapping(SLAM)functions to meet the requirements of working in this type of specific environment.The point cloud registration algorithm is an important part of the SLAM creation process.This thesis studies and analyzes the Iterative Closest Point(ICP)algorithm in the point cloud registration,using the AGV built in the laboratory as the platform,Optimized and improved the original point cloud registration problem.The effect of simulation analysis is good,and the point cloud fine registration is greatly improved.Starting from the basic hardware of the AGV,this thesis builds an AGV that can be used in medical services.First,it introduces the basic hardware of each part of the AGV,and analyzes the measurement of the single-line lidar.The distance and scanning principle,deduces the relevant formulas of point-to-point and point-to-line ICP algorithm in detail.After solving and recovering the two-dimensional laser point cloud,the point-to-point ICP algorithm and the point-to-line ICP algorithm were used to register and analyze the laser point clouds collected in three different environments.In a further comparison experiment,the source point cloud data was partially occluded and the registration effect of the original algorithm was observed.At this time,the original algorithm could no longer achieve the fine registration of the source point cloud and the target point cloud.In response to this situation,this paper proposes a variable threshold corresponding point to improve the algorithm.In the six registration simulation experiments before and after the same environment,the improved root mean square value of the corresponding point error is reduced by 7%,and the average calculation speed of the iterative algorithm is increased by at least 10%.The improved algorithm has smaller corresponding point errors and shorter algorithm time-consuming,this allows the medical service AGV to better cooperate with medical staff to complete related tasks.