Research On Autonomous Navigation System Of Mobile Robot Based On ROS

In recent years,with the increasing application of mobile robots in various fields of society,the key technologies associated with them have attracted close attention from relevant researchers in domestic and overseas.For the indoor environment,the autonomous navigation function is the basis for the mobile robot to achieve a variety of additional service functions,and it is an important embodiment of robot intelligence.In order to improve the development efficiency,this project studies the autonomous navigation of mobile robots based on ROS,including ROS-based robot integrated system design,robot pose estimation,grid map construction,global path planning,local path planning,etc.The main research contents are as follows:1.The ROS-based mobile robot integrated system hardware platform is set up to prepare for the validation of the autonomous navigation function algorithm.Firstly,this project introduces and analyzes the autonomous navigation framework of ROS,establishes the navigation system framework of this project.Then it conducts detailed research and analysis on the working principles and original errors of photoelectric encoder sensors,IMU and laser radar.Finally,the differential movement model is analyzed and the odometer model of this project is established.2.In order to overcome the problem that it is easy for the single sensor to generate large positioning error,the Extended Kalman Filter is used to effectively fuse the data of both IMU and odometer,so that the pose estimation accuracy of the mobile robot could be increased initially.The revised observation model is proposed as an important distribution function of RBPF-SLAM,combined with the external observation information of the laser radar,to further improve the robot's autonomous positioning accuracy and the grid map's accuracy.3.The project has effectively improved the original RRT* algorithm,and combined improved RRT* algorithm with DWA algorithm,and the hybrid algorithm of the two method is regarded as the path planning algorithm of the subject.Firstly,the project has introduced a target bias strategy to solve the problem of slow convergence rate of RRT* algorithm.At the same time,a “step extension avoiding collision method” is proposed to improve the obstacle avoidance characteristics of the robot.Then the “reverse order connection method” is used to smooth the global path of RRT* planning,so that the stability of robot's motion can be improved.Then the principle and evaluation function of the DWA algorithm are analyzed in detail.Finally,a hybrid algorithm combining the two method is tested and verified in Matlab.4.The specific experiment of the above improvement proposals is performed on an autonomous ROS experimental mobile platform in the actual application scenario.Firstly,the feasibility of the proposed improved RBPF-SLAM scheme is verified.Then,an autonomous navigation experiment under various obstacle scenarios is performed to verify the effectiveness of the improved path planning algorithm.