An agent-based approach to introductory robotics using robotic soccer

Introducing robotics to undergraduates is a challenging problem from an educational standpoint. The overlap to numerous areas of applied computer science, and the number of difficult subproblems that must be dealt with in any nontrivial robotics application (e.g., vision, path planning, real-time computing), easily overwhelm students at the undergraduate level. We have been employing robotic soccer as a vehicle to motivate undergraduate students in robotics, in order to provide students with an interesting domain that embodies significant research challenges in artificial intelligence. In order to manage the complexity of the domain, we employ software agents in several key components in our approach, and as a methodology for student implementations. This paper describes our strategies for introducing the elements of soccer playing to undergraduate robotics students, and the agent-based approach we employ.