A Practical A.I. Development Module for a Robotics Course

The cheap and powerful personal computer (PC) has become an important and efficient tool for supporting engineering education. In this paper a PC-based training module, AIROBOT,is presented. The purpose of this module is to provide a platform for students to develop and experiment with artificial intelligence techniques. The training module, AIROBOT,utilizes an electronic noughts and crosses game board which is interfaced to the Scorbot-ER VII robot and a PC. The development and implementation of the module are discussed. Two techniques developed by the students are presented to illustrate the utilization of the module. The first technique involves searching a game tree data structure. The learning involves the on-line generation of the game tree as the games are played. An evaluation function is used to facilitate the search. The other technique is based on the artificial neural network approach using the backpropagation paradigm. The structure of the neural networks, training and performances are presented. The PC-based training module has the potential to enhance the student understanding through the practical application of artificial intelligence. It is envisaged that similar modules can be easily integrated into most engineering undergraduate robotics courses.