Adaptive Robots
Adaptive Robots
There exist different methodologies for controlling robots. The aim of this course is to achieve a knowledge about these methodologies and their applicability under given circumstances. We will put emphasis on adaptive and autonomous robots that can be developed with neural network and evolutionary algorithm approaches. Other approaches, such as the behavior-based approach to robotics and traditional artificial intelligence approaches, will also be covered. Generally, we will investigate the motor-sensory mechanisms that engender intelligent behaviour in artificial autonomous systems, robots. In some cases, it might be feasible to use robot simulators --- this demands an exact transferring from the simulator to the real robot. In other cases, development of a controller must necessarily take place on a real robot. For investigating these issues, students will be divided into groups of 2-3, each group will be handed a robot (either a Khepera robot or a LEGO robot), and become able to develop appropriate controllers for their robot.
LecturerHenrik Hautop Lund
Literaturedistributed notes and articles.
Valentino Braitenberg's "Vehicles: Experiments in Synthetic Psychology",
MIT Press, Cambridge Massachusetts, 1984 (ISBN 0-262-52112-1 paper)