Research On Visual Location Method By Fusing Multi-line Lidar Data

With the development of sensor technology and control technology,robot navigation,virtual reality,and augmented reality(AR)have become research hotspots in industry and academia.Visual sensors are affordable in price and light in weight,and they are widely deployed on mobile devices such as mobile phones.Therefore,the locating method based on visual sensors has become a popular research direction.In practical applications,the point cloud map's scale information generated by the monocular camera is lacking.Simultaneously,the complex indoor environment will have a great impact on the operation of the visual locating method.Aiming at the above problem,this paper focuses on the visual locating algorithm in the complex indoor environment.The main work is as follows:(1)Investigate the domestic and international research development of visual SLAM,structure from motion(Sf M),and visual locating algorithm based on point cloud map.Take SIFT and ORB as examples to analyze the extraction and matching methods of feature points,and analyze 2D-2D,2D-3D,and 3D-3D motion solution method.(2)Study the principle and method of the monocular camera's internal parameter calibration and its external parameter calibration with multi-line lidar.Explain the realization method of multi-line lidar odometry.A point cloud map scale calculation method fused with multi-line lidar is proposed based on the trajectory output by the multi-line laser odometry and the external parameter calibration results.(3)Affected by the repetitive texture structure and moving objects in the indoor environment,the locating method based on the point cloud map is subject to locating errors and inaccurate locating.To tackle this problem,a visual positioning method based on the recognition map is proposed.Based on the image processing library Open CV and the optimization library Ceres,the related experiments are carried out in actual scenes,which prove the accuracy and robustness of the method in this paper.The end of the thesis provides conclusive remarks,explains the innovations and main results of this research,and points out the future research direction.