Research On SLAM And Autonomous Navigation Of Quadruped Inspection Robot Based On ROS

Recently,mobile robot technology has developed rapidly,and the demand for mobile robots in various fields is increasing day by day.The key problem of mobile robot is autonomous navigation technology,that is the function to realize autonomous movement of the robot in unknown environment with its own environmental perception device.Due to the limitation that wheeled robot is difficult to cross obstacles,this paper studies the autonomous navigation system of quadruped robot.Taking lidar as the main environmental perception device,an autonomous navigation system of quadruped robot is proposed,and relevant experiments are carried out both on ROS and A1 robot.Firstly,the software frame and hardware used in this paper are introduced.The robot chassis model,lidar model and environment model are established,and a complete design of navigation system of quadruped robot is proposed.Secondly,the simultaneous Localization and Mapping(SLAM)technology of mobile quadruped is studied.The grid map is selected as the representation of environmental in this paper.Aiming at the particle degradation problem of traditional RBPF-SLAM algorithm,an improved RBPF-SLAM algorithm is proposed.Then the graph-based SLAM algorithm is studied.Then the simulation environment is built in ROS and the simulation experiments are carried out for the two algorithms and the superiority of graph-based SLAM algorithm verified.Thirdly,the path planning technology of quadruped robot is studied.The basic principles and processes of global path planning algorithm,A* algorithm and Dijkstra algorithm,as well as local path planning algorithm,DWA algorithm and trajectory rollout algorithm,are introduced in detail.The simulation experiments of the above four algorithms are carried out in ROS system,the simulation results are analyzed,the advantages of various algorithms are obtained,and the algorithm interface is set to switch the path planning algorithm.Finally,the physical experiment environment is built by using the existing conditions of the laboratory.The A1 robot is used to test the improved RBPF-SLAM algorithm and the graph-based SLAM algorithm.After that,A* algorithm and DWA algorithm are combined as navigation algorithm group I,Dijkstra algorithm and trajectory rollout algorithm are combined as navigation algorithm group II.Then A1 robot is used for physical navigation experiment to test the two group of algorithms.By comparing and analyzing the results of the above algorithms in a series of experiments in environments concludes with complex environments,unknown and dynamic obstacles,a set of autonomous navigation system of quadruped robot is finally integrated,and the experimental results show that the system has high mapping accuracy and can be competent for navigation in complex environments including dynamic obstacles.