Map Construction And Optimization Of Mobile Robot Based On Laser SLAM

Since the concept of Simultaneous Localization and Mapping(SLAM)was first proposed in the 1980 s,the relevant theories of SLAM have become matured gradually,and have been widely used in mobile robots,unmanned driving,augmented reality and so on.Laser SLAM particularly has been widely used in the field of mobile robots.Although the theory of laser SLAM has been relatively mature,there are still some problems such as inaccurate positioning and map overlap when faced with the changing environments in practical applications.The main research of this thesis is the construction and optimization of map based on laser SLAM,focusing on solving the problem that the positioning accuracy of mobile robot is not high in practical applications.First,this paper analyzes the basic system framework of mobile robots,and designs the mobile robot prototype based on the research requirements.According to the robot's movement mode and the observation sensor the robot uses,the odometry model of the differential motion and the observation model of the Lidar are established,and the noise model of the Lidar is analyzed.According to the characteristics of data collected by Lidar,this paper analyzes the process of two kinds of scanning matching methods of Lidar and their advantages and disadvantages,and selects feature-based scanning matching as the method of Lidar positioning.Then,this paper proposes the method of feature extraction for natural features and artificial features.In view of the natural features,this paper mainly studies the extraction of line segment features.The method of line segment feature extraction is studied and then the appropriate extraction parameter is determined through experiments.For the artificial features,the form of the artificial features is first determined,and then two parameters for extracting artificial features are proposed.The thresholds of these two parameters are determined experimentally,and the detection range that the artificial features can be extracted effectively by Lidar is measured.Then,this paper conducts related research on the map construction of laser SLAM.In this paper,the two-dimensional occupancy grid map is used to describe the environment space of the robot.The update method of the occupying grid is proposed,and the environment map of the laser SLAM is constructed.In view of the global consistency of the map,this paper proposes to use the graph optimization method to establish constraints between the poses of the robots,between the natural features and between the artificial features,and then global optimization is conducted to obtain a map with global consistency.Finally,the algorithm performance of laser SLAM is verified by experiments.Let the robot perform navigation tasks using the environment map constructed by the laser SLAM and observe whether the robot can complete the navigation task.Test the positioning accuracy of the robot when it enters the station during navigation and take it as a basis for judging the positioning accuracy of the laser SLAM in order to evaluate whether the positioning accuracy of the laser SLAM can meet the actual requirements.