2D LiDAR SLAM Method Of Mapping Robots For Indoor Scenes Of Buildings

At present,the main expression of indoor environment of buildings is two-dimensional map.For robots,the acquisition of environmental map is mainly through simultaneous localization and mapping(SLAM)technology.Although 2D laser slam technology is relatively mature,but for complex,large-scale,similar features of indoor scenes,the traditional slam scheme has inaccurate location and rough map.This paper focuses on the application requirements of Surveying and mapping robot,based on the slam method of graphical model,through the robot modeling,odometer calibration,the improvement of SLAM algorithm,global map construction,and on the basis of laser slam theory,a set of positioning and mapping scheme for indoor scene of Surveying and mapping robot is developed,the main contents and conclusions are as follows:1.Based on the indoor 2D surveying and mapping requirements of Surveying and mapping robot,the robot motion model,sensor observation model and map model are analyzed and established.On this basis,the mathematical model of SLAM problem is described,and a unified platform is built for subsequent research.2.Aiming at the problem that the wheel odometer of mobile robot is easy to accumulate errors,a method of odometer calibration based on linear relationship is proposed.In the stage of laser truth acquisition,imls-icp algorithm is proposed to improve the convergence speed of the algorithm;in the stage of solving linear equations,the internal parameters of odometer calibration are determined by the least square method to reduce the system error caused by odometer.3.Aiming at the problem that the slam method of scan-to-map matching is easy to be interfered by the initial value,an improved scan-to-map matching is proposed.The robot's pose is obtained by Ceres scan matcher,and the submap is estimated by this method.This way of building global map with multi submap reduces the interference of the initial value,which improves the map accuracy by 44.1% compared with the ordinary slam method;In dealing with local extremum,a quadtree pruning algorithm is proposed,which improves the matching efficiency by 38.24%;In order to solve the problem of robot positioning,the closed-loop detection is introduced to correct the robot's position and pose,reducing the robot's translation error by 79.06%.4.Aiming at the low efficiency of traditional indoor surveying and mapping and the problem of human interference,a solid surveying and mapping robot platform is built,and the indoor space is measured and constructed with the improved scan-to-map matching slam technology and the grid map environment representation method.Compared with the traditional surveying and mapping methods,the reliability of the improved slam method in indoor surveying and mapping is proved.