Design Of Indoor Mobile Robot Positioning And Mapping System Based On Lidar

Real-time positioning and mapping is the basis for indoor mobile service robots to expand other businesses.Firstly,the robot needs sensors to sense the surrounding environment in order to complete autonomous movement.The integrated use of lidar and odometer is the mainstream direction of current applications.Therefore,the research of SLAM system based on this method is of great significance.In this paper,the selection of hardware platform and the design of software system are presented based on laser radar and odometer for indoor environment.Aiming at the lack of accuracy and robustness of the existing algorithm,an improved positioning mapping system was researched,designed and tested.The main research contents are as follows:First of all,this paper introduces the ranging principle of different types of lidar and the software and hardware framework of the system.Moreover,a simple and flexible calibration method is designed to calculate the coordinate transformation relationship between the odometer and the lidar,which provides a theoretical basis for the construction of the software algorithm.Secondly,the principles and defects of the traditional ICP and CSM algorithms are analyzed.The ICP is used to complete the framing and improve the quality of the point cloud.At the same time,the Gauss-Newton iterative optimization algorithm is introduced to improve the CSM algorithm and improve the positioning accuracy.According to the analysis of unstructured scenes,a covariance-based reliability evaluation scheme for fixed positions is proposed.Simulation experiments show that this scheme can effectively distinguish between structured and unstructured scenes.The penalty factor is also introduced to increase the weight of the motion model and reduce the weight of the observation model,so as to optimize the problem of easy positioning errors in unstructured scenes,and improve the robustness of traditional CSM positioning.Again,considering the cumulative error caused by long time positioning,the loop closure detection module is added.A key frame selection strategy is proposed in this module,which can eliminate redundant key frames while providing a sufficient number of effective key frames,and combines angle information to eliminate singular key frame point clouds.After that,the multi-resolution map and branch and bound method are also considered to optimize the relocation in the loop closure detection module.Experiments verify that the loopback detection has completed the optimization of the entire pose graph,which reduces the overall error and the time spent in constructing the map.Finally,the paper experimented the laser SLAM algorithm system constructed in this paper through the open source database set and analyzed the system error distribution of the traditional CSM algorithm,the optimized CSM algorithm in this paper,and the optimized algorithm after adding loop closure detection.The experimental results show that the average positioning error per unit length of the lidar positioning system designed in this paper is about 0.6% and 0.5% on the two public data sets,which is better than the 9% and 5.2% of the traditional CSM algorithm.The root mean square errors are 1.66 m and 1.24 m respectively,which are greatly improved compared to the 6.93 m and 8.94 m of the traditional CSM algorithm.Later,the system was tested in a real environment,and the results also showed that the system can provide more precise positioning and map information in more unstructured scenarios,and has better robustness.