Research On SLAM Technology And Path Planning Algorithms For Unmanned Vehicle

With the continuous development of intelligent unmanned vehicle technology,more and more unmanned vehicles of various forms have been applied in industry.In order to make people's life faster and more convenient,more researchers are devoted to the research in the field of intelligent cars.The exploration of unknown environments for mobile unmanned vehicles has always been a hot and difficult research topic,among which localization and map building(SLAM)and path planning are the key technologies for autonomous localization and navigation of unmanned vehicles.Firstly,starting from the localization and map construction(SLAM)of unmanned vehicles,this paper introduces the coordinate system commonly used in the movement process of unmanned vehicles,and establishes the circular motion model of mobile unmanned vehicles for subsequent algorithm analysis.In this paper,the extended Kalman filtering based SLAM algorithm,particle filtering based RBPF-SLAM algorithm,Gmapping algorithm and Cartographer algorithm are used to analyze the principles.The advantages and disadvantages of several mapping schemes are compared,and the simulation tests and analysis of different mapping algorithms are carried out.The basic mapping algorithm framework for autonomous navigation is selected.Path planning for mobile unmanned vehicles is one of the most important core technologies of unmanned vehicles.The planned path should be optimal and conflictfree as far as possible.This paper analyzes and compares the global path planning techniques based on Dijkstra and A star.A star algorithm introduces heuristic function to reduce the search of nodes,which not only retains the advantage of Dijkstra algorithm to get the shortest path,but also takes into account the advantage of breadthfirst search.Therefore,A star algorithm is selected as the global path planning in this topic.In the aspect of local path planning,the traditional DWA algorithm based on dynamic window is selected as the main algorithm.By introducing the evaluation function of global information,the problem of falling into local optimum is solved,while the performance advantage of real-time obstacle avoidance of the unmanned vehicle is retained.Finally,this paper takes the four-wheel Mecanum Wheel unmanned vehicle as the experimental platform,designs and realizes the autonomous navigation of the unmanned vehicle in the ROS system,combines the positioning and map building algorithm and path planning scheme studied in this paper,and carries out the full physical experiment in the indoor scene.The correctness and availability of the indoor location and map construction algorithm and path planning scheme are verified.