Research On Robot Navigation Based On Binocular Stereo Vision

This thesis based on binocular stero vision has studied the navigation of the mobile robot.Navigation of binocular stero vision is a kind of new navigation method,that rely on a couple of camera that fixed on the mobile robot get the information of environment,the 3D map that get from this information be send to navigation arithmetic to get best path to avoid obstacle ane lead robot move forward.Though there are mang navigation method,but for the mobile robot in unkown environment the binocular stero vision navigation is a better method. Compared to other approaches, stereo vision navigation is rapid, accurate and easy to conceal, as it is a passive measurement, which does not emit light and radiation. And it can get the balance point form real time performance and accuracy of navigation. At the present time,the main shortcome of stero vision technology is bad real time performance,too much time be spent at for-sequence dispose.and lack a way to rebuild very accurate 3D map.The navigation of binocular stero vision include binocular stero vision technology,map build,navigation algorithm three parts.The binocular stero vision technology include digital image obtain,camera calibration,image preprocessing,feature-base matching,depth map build parts.Map build include 3D rebuild and map expression.Navigation algorithm include path planning summarize and based on improved genetic algorithm path planning.There are two study emphases in this thesis.s: The first one is think over binocular stereo vision technology and the navigation as a whole consideration. Follow the purpose of achieve duty as the guidance, based on the feedback principle, binocular stero vision technology short-cut method which may measure to real time performance and the navigation accuracy. The second emphases is the map build. Proposed one in the depth chart, the binocular stero vision part's primitive image to with the line chart foundation, take the feature reverse retrieval match's thought as the instruction, according to by getting line, by the line getting from surface, by looks unfamiliar the adult, to transfer the 3D step by 2.5 dimension, the additional coordinate transformation and the wrong inspection function, completes the three-dimension map reconstruction finally the method. In addition has carried on the proof after these two method's feasibility, has given the corresponding main algorithm.