Design and implementation of a vision-based robotic air hockey table

Abstract

Air hockey has been an arcade favourite since the early 70's and has gained as much popularity to even schedule a yearly world championship. Unfortunately, this game requires a minimum of two players making it impossible for users who wish to train or play by themselves. To solve this issue, a robotic air hockey table capable of emulating a human player has to be created. In this paper, we have developed a computer vision system to detect and track the movements of the puck which is used to predict the final position of the puck in real time. Once the system predicts the final location of the puck, an electrical linear actuator capable of high speed and acceleration is activated which drives the mallet to block the goal. This paper will also discuss various problems encountered and possible future solutions to eradicate them.