Towards Smart Robots: Rock-Paper-Scissors Gaming versus Human Players.

Abstract

Abstract. In this project a human robot interaction system was developed in order to let people naturally play rock-paper-scissors games against a smart robotic opponent. The robot does not perform random choices, the system is able to analyze the previous rounds trying to forecast the next move. A Machine Learning algorithm based on Gaussian Mixture Model (GMM) allows us to increase the percentage of robot victories. This is a very important aspect in the natural interaction between human and robot, in fact, people do not like playing against “stupid” machines, while they are stimulated in confronting with a skilled opponent. 1