A ping-pong ball catching and juggling robot: A real-time framework for vision guided acting of an industrial robot arm

This paper investigates the possibilities of real-time control of standard industrial robot arms by the means of a ping-pong ball juggling system that is able to catch a ball thrown by a human and to keep the ball airborne for more than 30 min. The sensors are two industrial cameras operating at 60 Hz. We discuss the image processing and the controlling algorithms used. Emphasis is placed on comparing the widespread naive linear model and a novel physically correct model of the flight trajectory of the ball. While both models are sufficient for the juggling task, it is shown that only the non-linear model is able to predict the impact parameters sufficiently early and precise enough to solve the initial catching task. The performance of the complete systems is also documented in a supplementary video [8].