Research And Implementation Of V-SLAM Algorithm Based On Monocular Camera

The indoor positioning technologies such as Bluetooth,WIFI,Inertial Navigation System and Ultrasound can meet the needs of users to a certain extent.However,these methods still have shortcomings such as low positioning accuracy,complex technical conditions and high cost,and there are some obstacles in the popularization of application.At present,the Simultaneous Localization and Mapping(SLAM)technology is in the ascendant.In this paper,according to the advantages of camera in obtaining environmental information,combining SLAM technology,and use monocular camera to get data.In the front-end,the improved ORB algorithm based on data association for image feature detection and matching is introduced,the key frame closed-loop detection mechanism to reduce mapping error is introduced in the back-end,and evaluate the error of camera trajectory by the relative pose error(RPE).The difference assessment provides a new reference and thinking for indoor positioning methods.The specific research contents and achievements of this paper mainly include:(1)The improved ORB algorithm based on data association is used for image feature detection and matching.The experimental results show that the improved ORB algorithm based on data association is superior to traditional algorithms such as SIFT,SURF and BRISK.(2)The key frame closed-loop detection mechanism is introduced in the back-end to further reduce the drift error of mapping;evaluate the motion trajectory of the camera by the relative pose error(RPE);researched the basic principle and method of camera calibration,analyzed the imaging principle of monocular camera,and the monocular camera used in the experiment is calibrated by using MATLAB a2014 software.The factors affecting the calibration results are analyzed.(3)The simulation and real environment data experiments were carried out,and realized the camera trajectory tracking and environment map construction.The experimental results show that the proposed visual SLAM algorithm based on monocular camera can improve the experimental efficiency and accuracy.The relative pose error of camera trajectory can be improved to less than 10 cm,and the experimental efficiency can be improved by more than 20%.