A high school project on artificial intelligence in robotics

In this paper, we describe the development of an artificial neural network strategy for an industrial robot to play the game of Tic-Tac-Toe. This project was undertaken by two high school students during the university's technology and engineering research programme. The strategy is based on the feedforward multilayered neural networks with backpropagation of error training. The performance of the strategy is evaluated by its accomplishment against human opponents. The results indicate that the neural network strategy developed will almost always win if given the opportunity and at most draw if not. The neural network strategy developed has been successfully interfaced with a Scorbot-ER VII robot via an in-house designed electronic game board.