| Highly intelligent, highly flexible and highly efficient mobile robots gain importance with booming of technology, and more and more domains need involvement of mobile robots. Hence to develop independent mobile robots is of all-important significance. To cope with the problems of low real-timeliness, weak capacity of positioning and hindrance avoiding, etc., the paper combines current machine vision as the hotspot of research on mobile robots autonomous navigation to take the study on mobile robot positioning and hindrance avoiding system with binocular vision.Firstly, it makes a contrastive analysis on typical vision system of mobile robots, selects binocular vision system as the subject investigated, introduces related theories of binocular vision positioning, and focuses on image processing algorithm of hindrance detection. According to characteristics of binocular vision system, it proposes image processing algorithm flow of hindrance detection. In this image processing algorithm flow, local stereo matching algorithm is adopted according to system's requirements on real-timeliness of matching algorithm and on matching effect. And experiment and contrastive analysis on matching effect of local stereo matching algorithm based on different matching costs are conducted. Then the paper suggests translating information of hindrance depth into binary image utilizing image thresholding idea according to hindrance depth information obtained by local stereo matching of image to separate hindrance from background, then recognizes hindrance through outline search technique. Lastly positioning coordinates of hindrance are set according to obtained 3D information of hindrance.Secondly, in hindrance avoiding, kinetic analysis for mobile robots is conducted. Then the paper suggests straight-line motion and arc motion as the hindrance avoiding strategy of mobile robots, and analyzes the condition to be met by the two hindrance avoiding strategies. Lastly, algorithm flow for hindrance avoiding strategy is designed.Lastly, considering higher requirements on computing capacity and real-timeliness by mobile robots' vision positioning and hindrance avoiding system, the paper builds system hardware platform using combination of upper computer and lower computer. Based on building block design method, upper computer software uses Visual studio 2013, Opencv vision pool and DirectShow flow media development package to compile procedure for modules of image processing algorithm and hindrance avoiding algorithm. The lower computer software adopts KEIL MDK to compile procedure for modules of master controller and motor driver to complete software design of the overall system. Lastly the system feasibility is verified through experiment. |
PDF Full Text Request