Shuttlecock Detection and Fall Point Prediction using Neural Networks

The prediction of moving objects has always been an interesting area of research. Lots of prediction algorithms have been developed in the past to predict the trajectory of moving objects. This paper deals with the prediction of shuttlecock trajectory in two dimensions. This work is a part of the shuttle badminton playing robot that is under development in our laboratory. The focus of this work is to carry out detection and prediction processes using single camera input only. Based on the obtained results, the work will be extended to stereoscopic cameras. The first part of the work deals with the detection of the shuttlecock using deep learning techniques and the rest of the paper deals with the prediction of the shuttlecock using Neural Networks. The obtained prediction is around 80 percent accurate and can be convincingly used as the target point for the navigation of the robot.