| With the advancement of science and technology and social development,mobile robots have been widely used in military,industrial,and civilian fields.The autonomous navigation and path planning capabilities of mobile robots have become a research hotspot in recent years.It mainly includes the use of various sensors to realize real-time synchronous localization and map construction(Simultaneous Localization and Mapping,SLAM)of the surrounding environment,multi-sensor fusion algorithms,and path planning methods based on the requirements of building maps and tasks.Due to its own limitation,2D lidar can only scan obstacles in one plane,and cannot complete the detection of low obstacles and hollow obstacles,and cannot meet the needs of perception in certain situations.This project aims to build a self-moving robot system with complete autonomous navigation and path planning based on the integration of 2D lidar and binocular cameras.First,optimize the traditional RBPF-SLAM algorithm,and then improve the ORB-SLAM2 system framework according to the needs of autonomous navigation,and then A loose coupling method of laser SLAM and visual SLAM is studied,and finally the full coverage path planning algorithm is researched.The main research of this paper is as follows:(1)A robot experimental test platform based on a crawler chassis is built,and the robot platform is divided into three parts: the upper computer software,the lower computer software,and the UI visualization interface according to the needs of autonomous navigation and detection.The lower computer system is responsible for the robot.The motion control of the experimental test platform and the reception of data from internal sensors(IMU,position sensor)and communication with the host computer system,the host computer system receives data from external sensors(lidar,binocular camera)and runs upper-level algorithms such as full coverage path Planning algorithm,synchronous positioning and map construction algorithm,lidar motion distortion removal algorithm,sensor map fusion algorithm,odometer integration,etc.,and communicate with the lower computer system.The UI visualization interface is responsible for the display of various information and human-computer interaction.(2)Using the lidar odometer based on the PL-ICP algorithm to complete the wheel odometry calibration using the direct linear calibration method;using the odometer data auxiliary method to complete the removal of the lidar motion distortion;calibration based on Zhang Zhengyou's chessboard Method to complete the calibration of the binocular camera;based on the 6*6 April Tag code matrix calibration board to complete the joint calibration of the binocular camera and the lidar.(3)Aiming at the simultaneous positioning and map construction of single-line lidar,this paper studies and optimizes the RBPF-SLAM algorithm.On the basis of the traditional RBPFSLAM algorithm,the seagull optimization algorithm is used to improve the particle sampling process of RBPF-SLAM,which avoids the degradation of excellent particles to very noisy particles;for the sample poverty caused by simple resampling in the RBPF-SLAM algorithm To solve the problem,the MSV resampling method is used to optimize.Then use the algorithm proposed in this paper to conduct simulation experiments on the Intel Research Lab and ACES Building data sets.From the experimental results,it can be seen that after the algorithm is optimized,the corresponding effectiveness is significantly improved.In order to verify the actual effect of the algorithm,an indoor SLAM experiment was carried out using an experimental test platform.The experimental results show that the optimized algorithm can complete the task of building a map and obtain a map with better consistency with the actual environment.(4)For the problem of synchronous positioning and map construction of binocular vision,this paper conducts research and analysis on the ORB-SLAM2 system.Aiming at the shortcomings of the ORB-SLAM2 system that there is only a sparse point cloud map,an octree map thread is added to the original system.The construction of the octree map is realized,so it can be directly used in robot navigation and path planning.In order to improve the rate at which the ORB-SLAM2 system eliminates mismatches,this paper uses the PROSAC algorithm to replace the RANSAC algorithm in the original framework,which improves the real-time performance of the overall algorithm.(5)The full coverage path planning algorithm based on biostimulation neural network and the full coverage path planning algorithm based on neuron excitation are implemented on MATLAB and compared and analyzed,and finally the better performance is selected based on the full coverage path based on neuron excitation The planning algorithm is used as the path planning algorithm of this subject. |
PDF Full Text Request