Research Of Mapping Method Based On ROS And RGBD Sensor

The mobile robot industry was born in the late 1960 s.In recent years,with the rapid development of high-tech science and technology,the research of mobile robots is more and more in-depth,and the application field is increasingly extensive.In the absence of prior knowledge,the process of simultaneous localization and mapping(SLAM)for mobile robots is an important research area,and the research of visual SLAM is an important research orientation in this field.On the premise of using RGBD vision sensor,this paper proposes an improved visual SLAM algorithm for mobile robots based on ROS.The research contents are mainly included into three parts:(1)The design and implementation of feature extraction and matching in indoor environment is the premise of realizing visual SLAM system.In indoor environment,RGBD camera collects color and depth information of image,and the system extracts and matches features according to image information.In this paper,three feature matching algorithms,SIFT,SURF and ORB,are compared in real-time,and then the initial matching and RANSAC matching are carried out according to the extracted images,in order to prepare the system for building a global map.(2)The design and implementation of the system based on RGBD-SLAM algorithm.Firstly,the system framework of visual SLAM is analyzed.Front-end,back-end optimization,mapping and loop detection are described.Back-end optimization based on BA,point cloud mosaic,loop detection and graph optimization based on g2 o are introduced.The accuracy and reliability of the RGBD-SLAM algorithm proposed in this paper are verified by constructing local three-dimensional point clouds in indoor environment with ICP algorithm and adding loop detection mechanism.(3)Design and implementation of robot map construction based on ROS and RBGD sensors.Firstly,the platform of ROS is built,and the communication configuration between the robot and PC is completed.In the experiment part,firstly,the RANSAC algorithm is used for feature matching.The experiment proves that the efficiency of the system in the feature matching stage is improved after the RANSAC algorithm is matched.Then,experiments are carried out on the ROS-based map building system of mobile robots to verify the feasibility of RGBD-SLAM algorithm,and the accuracy of map building is improved by adding loop detection mechanism.The experimental results show that the initial matching efficiency of the front-end part of visual SLAM is 63.46%,and after the second matching of RANSAC,the matching efficiency is 88.70%,which improves the system efficiency by 25.24% in the feature matching stage.Visual SLAM improves the mapping accuracy by 55.56% by adding loop detection mechanism.Although the details of the final three-dimensional map are somewhat blurred,the overall outline is basically clear and complete,which can meet the daily needs.