Design And Implementation Of Global Positioning Algorithm For Mobile Robot Based On Vision

Mobile robot positioning is the key to complete autonomous navigation for mobile robot.It is of great and realistic significance to carry out research on mobile robot positioning.The global visual positioning method is widely used in the multi-mobile robot with the advantage such as cost-effective and so on.However,in the process of global visual positioning,the algorithm is hard to get the mobile robot pose angle and poor real-time.In order to solve the above problems,the global positioning algorithm of mobile robot based on vision is designed and realized in this paper.It is mainly from the following aspects as follows:Firstly,the global visual positioning principle used in the experiment is analyzed.The internal parameters and distortion parameters of camera used in the experiment are obtained by calibrating the camera.Then the relationship between the image pixel coordinate system and the world coordinate system is established.And the camera image was corrected by using the obtained parameters.Studies on above works can provide adequate experimental basis for the subsequent completion of the global mobile robot vision localization experiment.Secondly,a Camshift combined with SIFT localization algorithm is proposed to solve the problems which mobile robot pose angle is hardly to get and mobile robot can't automatic recovery when the positioning is failure by Camshift.This paper uses Camshift to solve the mobile robot position coordinates,and then uses the orientation of keypoints to get the mobile robot pose angle such that the mobile robot pose information is obtained.When the Camshift algorithm fails in locating the mobile robot,the SIFT algorithm is used to locate and initialize the mobile robot,and then the mobile robot can be tracked by Camshift.The experiment studies are provided to demonstrate the validity of the proposed combined algorithm.Finally,aiming at the poor real-time problem of Camshift combined with SIFT algorithm,a method of locating mobile robot using BRISK is proposed.The BRISK keypoints descriptor extraction and matching are carried out on the mobile robot template and the current video image.Then the invariant keypoints topology is applied to filter the erroneous keypoints,which solves the problem of erroneous keypoints is hard to avoid.The keypoints vector angle is used to solve the pose angle of the mobile robot,and the keypoints are used to locate the mobile robot.The experiment studies are given to verify the effectiveness of the proposed algorithm.