A learning approach to robotic table tennis

We propose a method of controlling a table tennis robot so as to return the incoming ball to a desired point on the table with specified flight duration. The proposed method consists of the following three input-output maps implemented by means of locally weighted regression: 1) a map for predicting the impact time of the ball hit by the paddle and the ball position and velocity at that moment according to input vectors describing the state of the incoming ball; 2) a map representing a change in ball velocities before and after impact; and 3) a map giving the relation between the ball velocity just after impact and the landing point and time of the returned ball. We also propose a feed-forward control scheme based on iterative learning control to accurately achieve the stroke movement of the paddle as determined by using these maps. Experimental results including rallies with a human opponent are also reported to demonstrate the effectiveness of our approach.

I. Introduction

To perform tasks during intermittent interactions between robot and environment, the robot must be able to adjust the strength and timing of interactions during execution of the tasks. These are called hybrid (mixed continuous and discrete) control problems by Burridge et al., who gave hopping, catching, hitting, and juggling as typical examples [1]. These tasks have attracted the attention of researchers also in the area of experimental psychology [2]. However, it is difficult to find general approaches that are sufficiently tractable for the robot to perform these tasks. In this paper, we focus on the table tennis task that involves the intermittent nature of the robot–ball interaction in order to explore the hybrid control problems.