Research And Implementation Of Laser Radar Based Mobile Robot Navigation System

Mobile robots are gradually entering people's lives with the progressing of society and the rapid development of science and technology,and they are used widely in industry,business,and universe exploring,etc.As one of the core functions of mobile robots,autonomous navigation has become a research hotspot in the field of robots.This article aims at implementing a laser radar based mobile robot navigation system on ARM platform,and researches on several aspects such as the design and implementation of navigation system,map construction algorithm and path planning algorithm are made in this article.The main research content includes the following parts:(1)The overall plan of implementing a laser radar based navigation system on Cortex-A53 platform is designed based on the researches on key problems and relating technologies in mobile robot navigation system,and detailed steps of building the mobile robot navigation system on this platform is described.Basing on the study of SLAM algorithm that is used for map construction,a map with smoother edge is built by setting a proper threshold for scan matching score.(2)Researches on global path planning algorithm A* and its two improved algorithms MEA* and RA* are made: at first A* algorithm is studied in two aspects,the effects of three heuristics and several data structures on the efficiency of A* are compared,and Octile heuristic and a hybrid data structure is selected to optimize the A* algorithm.Then aiming at the problem that MEA* algorithm can't find path when there are no available neighbor nodes for minimum node,the MEA*-R algorithm is proposed,and in the experiment where there are 2200 scenes MEA* algorithm can find path in 141 scenes,while MEA*-R algorithm can solve the defects of MEA* algorithm,which can find path in all scenes.At last,the RA*-FE algorithm is proposed combining the RA* algorithm and the fast expansion strategy,whose path-finding time is 2.31% less than the RA* algorithm.Experiments show that the path-finding time of the MEA*-R algorithm and the RA*-FE algorithm is 50% less than the optimized A* algorithm.(3)Improved path planning algorithms are applied to the mobile robot,and the memory overhead of the RA*-FE algorithm is reduced by improving data structure,making it suitable for ARM platform with limited computing resources.Through experiments on mobile robot,the DWA algorithm that is more efficient than the Trajectory Rollout algorithm is chosen to plan local path for the robot.The MEA*-R algorithm and the RA*-FE algorithm were tested in the actual environment,compared with the default global path planning package navfn in the ROS navigation stack,the path-finding time of MEA*-R algorithm was reduced by 81.77%,and the RA*-FE algorithm's path-finding time was reduced by 77.11%.Finally,the obstacle avoidance tests and navigation system accuracy tests of the mobile robot are completed,and the mobile robot built in this paper can achieve static obstacle avoidance and dynamic obstacle avoidance,and the positioning error of the mobile robot is about 3.0~10.0cm,which verifies the feasibility and effectiveness of the laser radar based navigation system.