Research on robot catching ball based on binocular vision positioning

Abstract

In this paper, the robot used to catch the ball is studied, and a faster and more accurate detection and recognition method for high-speed moving objects is proposed, which can automatically judge the position of the ball and output the position information through TCP communication protocol. Based on the binocular vision system, this method identifies the small ball by calibrating the camera, processing the gray-scale image, transforming the image coordinates to spatial coordinates and Hough transform, and then realizes the spatial position and distance detection of the small ball. In addition, the ball receiving system and the robot platform are studied, so that they can be connected in real time and stably. It ensures the real-time and accuracy of the vision system to track the ball and meets the design requirements of the ball-catching robot.