Dynamics parametrically controlled by image correlations organize robot navigation

摘要: Behavior-based robot designs confront the problem how different elementary behaviors can be integrated. We address two aspects of this problem, stabilisation decisions based on changing behavioral requirements and fusion multiple sources qualitative sensory information. These issues are studied in context a vision-guided mobile that is endowed with ability to reach goal while it avoids obstacles. Behavior organized from “inside” robot. Even absense external stimuli internal dynamics generate behavior. By exploiting image correlations visual sensors provide coarse estimates spatial relations. immediately coupled into neural realized by fields obstacle avoidance homing an autonomous The background field approach theoretical work function cerebral cortex ([1]). Neural represent state space estimation control. Convergent information cooperates divergent competes shaping stable attractor states. states dynamical systems one hand instrumental for control behavior other they concise hypotheses interpretation input. concept dynamic deals effectively contradictory assures only behaviorally relevant extracted navigation scheme works succesfully real-time resolutions as poor 322 pixels.