Autonomous robots acting in complex, dynamic every-day environments must constantly adapt their abilities to their environment by means of learning during the execution of their tasks. In this thesis a robot control language is introduced, which allows to describe declaratively and execute complete learning processes in the program. This process includes the identification and recording of experiences, the learning process itself, and the integration of the learning result into the program. Thereby, learning becomes an integral part of the robot behavior. The language supports a wide range of learning problems and learning algorithms. The implementation is based on the definition of an experience as a compact summary of a problem solving episode by means of hybrid automata.
«
Autonomous robots acting in complex, dynamic every-day environments must constantly adapt their abilities to their environment by means of learning during the execution of their tasks. In this thesis a robot control language is introduced, which allows to describe declaratively and execute complete learning processes in the program. This process includes the identification and recording of experiences, the learning process itself, and the integration of the learning result into the program. Ther...
»