Study On Visual Navigation And Obstacle Avoidance Of Wheeled Mobile Robot

Visual navigation following a guideline is a typical one of the numerous applications of mobile robot. In preparing this thesis, the AS-R robot, with its two wheels differentially driven, was taken as the control object as well as the experimental platform on which the study on visual navigation and obstacle avoidance were performed.Visual navigation is composed of image processing and motion control:1. In the module of image processing, monocular camera was used to capture image and to access environmental information, and HSV color space model was used to extract path. By transformation between image and world coordinates the effects caused by image distortion could be eliminated, and the parameters of the path might be obtained with straight line segments fitted by using least squares.2. In the module of motion control, dynamic sliding mode controller and optimal controller were designed on the basis of kinematics model of the mobile robot to achieve accurate path tracking.Visual C++ integrated development environment was used to implement functions in each module. Experimental results showed that image processing module had good real-time performance and could extract the path parameters robustly. Controller module had led the robot following the guideline accurately.Obstacle avoidance is a basic function for autonomous mobile robot. For the problems of "potential field traps" and the unreachable target too close to obstacle, this thesis used an improved potential field method. Simulation results verified that this method could effectively avoid the obstacles and approach the target.