In this article we describe a project for an undergraduate artificial intelligence class. The project teaches neural networks using LEGO[R] handy board robots. Students construct robots with two motors and two photosensors. Photosensors provide readings that act as inputs for the neural network. Output values power the motors and maintain the robot along the designated path. In doing this project, students come to realize the difference between training a neural network and the trained neural network. The fun factor associated with this project has encouraged students to elect artificial intelligence as part of their course of study.