Research On Binocular Vision Ranging And Aiming Attitude Solving Algorithms For Mobile Robots

In recent years,with the continuous improvement of hardware computing power and the continuous breakthrough of computer vision technology,computer vision technology has been more and more widely used in embedded devices.Taking the RoboMaster Robotics Competiton hosted by DJI as the background,this paper mainly studies the problem of target detection?tracking and binocular positioning of the ground mobile robot in the competition,and makes mathematical modeling of the two axis platform mechanism on the mobile robot,through which the position information of the target is transformed into the rotation angle information of each axis of the platform.It includes the following contents:First of all,according to the characteristics of the target,this paper establishes the realization steps of the binocular vision system,determines the design indicators of the binocular vision system,and constructs the binocular target positioning accuracy analysis model.Secondly,the principle of triangulation geometric measurement of binocular vision is analyzed and explained;how to set the binocular camera for binocular camera is introduced,and the binocular camera calibration and binocular image correction are realized based on OpenCV.Next,this paper focuses on the research of stereo matching algorithm:the four steps of stereo matching algorithm(cost calculation,cost aggregation,disparity calculation,disparity refinement)are analyzed in depth;the characteristics and implementation process of stereo matching algorithm CBCA and CSCA are studied and analyzed.The stereo matching algorithm BM?SGBM?CBC A and CSC A is tested and compared in Teddy?recycle and artl data of Middlebury test set.The performance of matching algorithms shows that the disparity estimation accuracy of CBCA and CSCA are equal,which both higher than that of BM and SGBM,but the processing speed of CBCA algorithm is faster than that of CSCA algorithm;CBCA algorithm is used as a stereo matching algorithm for binocular ranging experiment,the results show that the measurement error within 3m range is less than 35mm,within 5m range measurement error is less than 50mm.Finally,the binocular vision target detection?tracking and positioning(stereo matching algorithm is CBCA)are realized on TX1.The two axis platform is modeled mathematically,and the position information of the target point is transformed into the angle information of each axis of the platform through the mathematical model.