Optimization And Implementation Based On RBPF-SLAM Algorithm

Simultaneous Localization and Mapping(SLAM)has been a hotspot and a difficult point in the research of autonomous mobile robot technology.In order to solve the SLAM problem,an external sensor is needed to realize the state representation of the robot to the surrounding environment.Commonly used sensors include laser radar and vision sensors.Based on Lidar's Simultaneous Localization and Mapping(SLAM)has achieved good development in indoor environment due to its high precision,strong anti-interference and wide detection range.This thesis mainly studies the lidar-based SLAM to solve indoor positioning and mapping problems.The main work of the thesis is to analyze the traditional RBPF-SLAM algorithm based on Rao-Blackwelled Particle Filter(RBPF)method,and propose the corresponding improvement methods for the existing scanning matching accuracy and the lack of particle diversity.The first,according to the characteristics of the edge points of obstacles,a scanning matching method based on feature points is proposed.The edge points are recorded as feature points,and the remaining data points are common points.The feature points occupy higher weights at the time of matching,which improving the accuracy of the robot positioning.The second,the adaptive partition resampling method is used instead of the importance resampling in the original algorithm.The particles are classified according to the weight,and only the particles with larger weights and smaller ones are resampled,while the particles with medium weights directly enter the pose estimation process of the next stage,which improves the diversity of the particles and alleviates the problem of particle shortage.The third,in the mobile robot verification,the distributed framework of the Robot Operating System(ROS)is used to realize the remote communication between the Raspberry Pi and the PC,and the effect of the construction is improved by rationally distributing the running platform of the algorithm.This thesis compares the result on the public dataset of Intel Research Lab and ACES Building.The results show that the optimized algorithm improves the positioning accuracy of the robot by 15%,and the number of resampling times in the whole environment is reduced by 20%.Then this thesis builds a mobile robot experimental platform,and implements the RBPF-SLAM algorithm on the platform.Finally,the laboratory environment and the corridor environment are tested to verify the effectiveness of the algorithm in practical applications.