System design for autonomous table tennis ball collecting robot

Abstract

This paper presents design methodologies and an implementation of autonomous table tennis ball collecting robot. The robot is designed with three major emphases to collect balls effectively in a real table tennis court: ball detection, navigation, and ball collection. For ball detection, a combination of blob detection and cascade classifier was utilized. The robot generates an optimal path toward detected balls using A* graph search algorithm and follows the path based on a mecanum-wheeled platform. The detected balls are collected using a suction motor and a 2-DOF active nozzle. The integrated system successfully demonstrates the given task in a real table tennis court.