Motivation Dynamics for Autonomous Composition of Navigation tasks

In this article, we physically demonstrate a reactive sensorimotor architecture for mobile robots whose behaviors are generated by motivation dynamics. Motivation dynamics uses continuous dynamical system to reactively compose low-level control vector fields using valuation functions that capture the potentially competing influences of external stimuli relative system's own internal state. We show naturally accommodates through standard signal processing tools, and can effectively encode repetitive higher level task composing several controllers achieve limit cycle in which robot repeatedly navigates toward two alternatively valuable goal locations commensurately alternating order. these robust perturbations, including imperfect models kinematics, sensor noise, disturbances, resulting from need traverse difficult terrain. argue provide useful alternative based on hybrid automata situations where operates at low close physical hardware.