| Navigation is an important part of mobile robot research. It refers that the robot perceives the surrounding environment and the condition of the mobile robot itself with sensors. Under the control of the module the robot moves from starting point to destination points. In recent years, visual navigation has become one of the hot topics in the study of the navigation.In this paper we present a visual navigation method without maps. We study the methods of image processing and robot control in the process of mobile robot visual navigation. With the aid of MATLAB and VC + + software,we perform the experiments and simulation.(1) Firstly we should know the navigation path. In this paper we study the image preprocessing methods and detect road edge by adopting canny method. We adopt Hough transform method for extracting the road border. By adopting mentioned methods we can identify the navigation path.(2) Secondly we carry out the camera calibration experiment. Based on the imaging model of camera and distortion model, we derive the relationship among the coordinate systems. Under the calibration theory of Zhang Zhengyou we carry out calibration experiments, making the image of the two-dimensional information convert to three-dimensional space information.(3) Thirdly, we research on mobile robot vision navigation. According to the different path of mobile robot, the navigation path can be divided into straight path and curve path of navigation. In the straight path navigation, this paper put forward the Hough transform to extract the road boundary using the vanishing point to complete the straight path navigation. Because the vanishing point does not exist in the curve navigation path, we puts forward the method of least square fitting path for navigation.(4) Fourthly we study the method of identifying obstacles. Based on the theory of super pixel we adopt it for identifying obstacles quickly. According to different color property between barriers and road, we make the image segmentation to multiple pixels. The result of experiments shows that mentioned method can identify obstacles quickly.(5) Lastly we study the methods of mobile robot control: On the basis of the mobile robot kinematics model, we design the inside and outside two layer control methods for mobile robot. Outer control is established on the basis of the visual sensor. Through processing image for mobile robot visual navigation parameters, the mobile robot modifies the location itself, modifying the navigation parameter again, until it reach the desired position. The inner control adopted fuzzy control method after image processing. Navigation parameters are used as input, the desired speed as output. For DC motor speed control, This paper established the mathematical model of the motor. PID and fuzzy PID controller are designed. By using simulink toolbox in Matlab we make the simulation, the simulation results show that the fuzzy PID control effect is better than PID controller. |
PDF Full Text Request