Research On The Dynamic Path Planing Of Wheeled Mobile Robot Based On Binocular Vision

The wheeled mobile robot has been applied in various kinds of field like industry, military and domestic use, since the rapid development of robot technology. As a consequence, the obstacle avoidance of mobile robot has become a hot research field. This thesis will propose a wheeled mobile robot system, with a differential-wheel-structure, based on the technology of binocular stereo vision, which can automatically avoid the avoidance in a dynamic environment.The first chapter introduces the research and application status of mobile robots, introduces representative mobile robot products, and emphatically clarified the basic concepts and research status of binocular stereo vision technology as well as obstacle avoidance techniques for mobile robots. Investigation of the existing algorithms for mobile robot obstacle avoidance is also done to make related analysis. The final part statesthe significance and content of the subject.The second chapter introduces the imagingprinciple of binocular stereo vision system to clarify the linear perspective projection camera model and the presence of non-linear distortion. It uses checkerboard method to perform camera calibration experiments. After getting internal and external parameters of the camera, this paper does nonlinear optimization on the left and right camera’s calibration results, to give a more accurate binocular camera imaging model.The third chapter introduces the basic principles and methods of stereo matching, and then introduces the concepts of the pole line correction and constraint matching. This chapter also uses Bouguet algorithm in OpenCV platform for correcting image’s pole line, and uses SIFT algorithm for three-dimensionally matching binocular images.The fourth chapter, focusing on the problem that the calculation of the whole reconstruction for the three-dimensional environment is excessively large, makes research on the problem when identifying and tracking the feature obstacles points, and does experiments. Then it introduces the basic concepts of three-dimensional reconstruction. It also uses the least squares method for progressing a more accurate three-dimensional reconstruction experiments, which helps the study of measuring the obstacles distance and speed for binocular vision system.The fifth chapter introduces the basic principles and existing limitations of traditional artificial potential field.In terms of obstacle avoidance in dynamic environment, this chapter proposes to introduce speed into the construction of artificial potential field, and solves the problems of regional minima traps and that target cannot be reached.Finally, it shows the simulation experiment results of the improved artificialpotential field, which is made under Matlab platform.The sixth chapter introduces the hardware architecture and software architecture of the wheeled mobile robot platform used in this study, and does analysis on the differential driven structure of wheeled mobile robot. Practically, the paper shows the construction of a regional obstacle avoidance map for robots, where obstacle environment is constructed, and experimental verification of dynamic obstacle avoidance is done to prove the results.