The Vision Orientation System Based on OpenCV of the Glass Curtain Wall Cleaning Robot

This paper proposes a vision orientation system based on OpenCV to solve the orientation problem that glass curtain wall cleaning robots meet with at work. The system applies edge detection and the Hough transform to the images captured by the monocular camera, then judges whether the robot has reached glass edges according to the results of histogram comparison between the target image and the template image. Finally the system sends corresponding orders to the robot for path selection and orientation.