Robot Control for Dynamic Environment Using Vision and Autocalibration

To enhance flexibility and extend the area of applications for robotic systems, it is important that the systems are capable ofhandling uncertainties and respond to (random) human behaviour.A vision systemmust very often be able to work in a dynamical ``noisy'' world where theplacement ofobjects can vary within certain restrictions. Furthermore it would be useful ifthe system is able to recover automatically after serious changes have beenapplied, for instance if the camera has been moved.In this paper an implementationof such a system is described. The system is a robotcapable of playing checkers with a human opponent using a camera to determine wherethe board and pieces are located and when the human has made his move. The paperprimarily focuses on the vision aspect of the implemented system.