Mobile Robot Vision Location Research Based On Improved Camera Calibration

The mobile robot localization is an important topic in the research and applicationfield of the robot. It is the premise of the robot intelligent. This paper presents animproved direct linear calibration method in the mobile robot vision location. The methodcollects ground information to obtain the natural landmark using a camera on the robot.Firstly, the target image uses the image processing technology, obtains the pixelcoordinates of the target point in the image coordinate system, then uses the improveddirect linear calibration method to obtains the target point of the actual coordinate system,finally uses the relative geometrical relationship of the robot and the feature points to getthe position of the robot, and processes mobile robot vision location.Firstly, this article compares the effect of Roberts operator, Sobel operator, Prewittoperator and Canny operator. According the real-time requirements of the mobile robot,the experiments select the Sobel edge detection operator. For the local maximum ofHough Transform uses the improved Hough transform process to fit the line. The featurepoints are wiped from the line of the image. In addition, it is improve the accuracy of theline.Secondly, this paper presents an improved direct linear calibration method. Don’tmove the camera, the objects of calibration selects a planar template to simplify themapping relations of the image coordinate system and the world coordinate system. Wesolve the inside and outside parameters of the camera through the characteristics of animage pointed, so get the actual coordinates of the characteristic point. The speed of thecalculation is fast, it avoids the tedious of nonlinear optimization and the instability of thesystem, so the mathod meets the practicality of the camera calibration.Finally, experiments verify the correct and feasibility of above methods in Mat lab7.5 environment. Using the relative geometric relationship of the robot and the piontsmatches the continuous acquisition of two images match, and gets the position of the robot,this basically realizes the mobile robot localization in the laboratory, and analyzes thepositioning errors.