The system of video-data processing for the autonomous control of mobile robot

摘要

We describe the system of video-data processing based on monocular vision for autonomous control of mobile robot. The system allows detecting obstacles in a robot environment modelled as a set of straight-line segment. The given system consists of three basic stages and uses original algorithms, ensuring the required precision and realization of the system in the real time. The first stage uses a fast edge detection algorithm on the basis of two-dimensional Walsh transform. The algorithm of modified Hough transform is used for detection of straight-line segments. The third stage "segment tracking" uses Kalman filtration for tracking segments in a monocular sequence of images.