Design and implementation of a vision-based basketball shooting robot

By combining technologies of computer vision [(L.G. Shapiro et al., 2001),(R. Jain et al., 1995),(S. Umbaugh, 1998),(D.A. Forsyth et al., 2003)] and control systems, we designed a vision-based basketball shooting robot. It is a wheeled mobile robot equipped with a Web camera and a shooting arm. Using its vision, mobility and special mechanism to adjust shooting angles, the robot can aim for the target and shoot the ball into the basket autonomously. In addition, we proposed a visual feedback method [(W.K. Pratt, 2001),(R.C. Gonzales et al., 2002)] which uses laser pointers to spot and then modify the shooting angles to improve shooting precision. Besides, the shooting robot is trained with artificial neural networks [S. Haykin 1999] in order to build the learning capability for both memorizing the training data and generalization.