The goalkeeper strategy of robocup MSL based on dual image source

Abstract

Gatekeeping is one of the most important tasks in MSL, especially defending aerial shots. This paper proposes a goalkeeper strategy based on combining two image sources. The goalkeeper uses the Omni-vision system to adjust heading orientation of the Kinect sensor for capturing both depth and RGB images of the ball. The depth information is used to detect the spatial position of the ball and the RGB image ensures the validity of the recognition. Once opponents shoot, the proposed system uses the first few detected positions of the ball to predict the interception point by the goalkeeper based on the least square method. The accuracy and effectiveness of the proposed strategy are verified by experimentation.