A laboratory course in behavior-based robotics

Robotics has received a great deal of attention as a vehicle for motivating students to learn artificial intelligence, computer science, and even general science and engineering. Many courses involve the construction and programming of simple robots. Although these courses are excellent vehicles for teaching general engineering concepts, especially electrical and mechanical engineering, they tend to be dominated by hardware concerns. They cannot address what AI researchers think are the core concerns in designing autonomous robots-representation, planning, sensor fusion, and so on. This is because the robots lack the sensor suites needed to make these problems practical. At Northwestern University, we have taught a laboratory course that uses research-grade mobile robots to explore concepts of autonomy, such as sensor interpretation, planning, and action selection. The course uses low-cost, state-of-the-art hardware that provides not only sonar and odometric sensing but also several real-time vision systems. The author describes the course's impact on the students' learning experience and their ability to creatively select goals and approaches.