A Real-time Stereo Visual Servoing for Moving Object Grasping Based Parallel Algorithms

This paper proposes a visual servo system for real-time tracking and grasping of a moving object. In the visual servo system, a fast template matching method can be used in combination with a fast and reliable stereo matching method to achieve image matching of a moving object. Kalman filtering is then applied to provide improved position and speed estimation. At the same time, a parallel method is adopted to raise matching speed of the image matching method. The result is transmitted directly to robot controller and stereo-visual platform. It achieves the visual servoing of the robot and realizes real-time moving object grasping. By carrying out computer simulations and many experiments of object tracking with the capture of a moving object, experimental data is obtained to show that both oriented precision and matching speed have improved.