Homeokinetic approach to autonomous learning in mobile robots

We study robotic systems under the dynamical system paradigm aiming at a systematic approach to the self-organization of behavior. We introduce a new principle - homeokinesis - which is completely unspecific and yet induces specific, seemingly goal-oriented behaviors of an agent in a complex external world. The agent is equipped with an adaptive model of its behavior. A learning signal for both the model and the controller is derived from the misfit between model and true behavior. Using retrospective models the emerging behaviors are found to be both explorative and dynamically situated.