Mobile Robot Localization Design Based On Visual Information

Accurate positioning for the mobile robot itself is a basic and important function in mobile robot navigation, attracting great attention around the world. A summary of technologies in recognition, positioning is made as preparation and on the basis of which classic image edge detection methods, linear filtering technology and its application in image edge detection are analyzed firstly. Then various algorithms are implemented in edge detection to draw a comparison through theory analysis and simulation. At last, a mobile robot positioning system utilizing color-based image segmentation technology is proposed. Color camera and image acquisition card are used to acquire video image which are then converted into digital signal for computer procession. Image processing program is run to realize robot tracking and positioning, the results of which are displayed in the computer in real time.Threshold segmentation technology based on color features which include hue component H and saturation component S of HSI color space is adopted in robot positioning. Robot position is determined by twice searching method and predicted by prior location information to improve precision. Information for every two pixel points is compared with the set threshold to improve real-time ability. Considering perspective distortion occurring in the images taken by the camera, a kind of camera calibration method based on geometric mapping relationship is put forward to convert the robot position information under image coordinate system into real coordinate system.Software part of the system is composed of two components which are image processing software and application systematic software based on image acquisition card application interface library. The former is to analyze the performance of target testing image while the latter is to dispatch the overall system's software and hardware to realize positioning and results display for mobile robot.