A very short story about autonomous robots

G. Steinbauer, R. Koholka, and W. Maass

Abstract:

Machine learning for autonomous mobile robots is a very interesting but also very difficult problem. In this domain you need robust, fast and efficient learning algorithms. The combination of machine learning and mobile robots allows us to create machines whose performance surpasses both explicitly programmed robots and humans. Although humans can learn, their sensors and actuators are in several aspects inferior to those of robots. On the other hand a non-learning robot can only perform tasks where all the individual difficulties and complexities can be anticipated by the programmer of the robot. In this short article we discuss both a completed project, and a new one that has just started. The first project produced a robot that applied learning to improve his minigolf skills. The new project is the creation of the first Austrian team of autonomous robot soccer players that will compete in the robot soccer world championship RoboCup1.

Reference: G. Steinbauer, R. Koholka, and W. Maass. A very short story about autonomous robots. Special Issue on Foundations of Information Processing of TELEMATIK, 8(1):26-29, 2002.