Research Accomplishments and Interests

If autonomous systems are to realize their potential, they must improve their abilities to operate in the real world. Sensorimotor systems are among the weakest and most critical components of such systems. Perceptual systems exist to enable their owners to cope with the dynamic environments in which they operate. The system is inundated by stimuli; the successful system selects the stimuli that are critically relevant to its current activities and responds appropriately to them. My work is sometimes inspired by principles observed in biological creatures and sensorimotor systems that have stood the test of time. The gaze control work was motivated by the desire to design indexical robot behaviors, which are performed relative to a fixation point or other reference point. The capability for a robot to go-to-the-fixated-object yields robust navigation based on visual feedback despite errors in other senses (e.g., slippage errors in dead-reckoning). My work on low-resolution peripheral vision for mobility was inspired by the honeybee’s use of peripheral optical flow to center its flight path through a narrow gap. Reflexive visuomotor behaviors alone cannot achieve our expectations for intelligent robot applications. I am investigating the robot’s use of visual attention in building its representations and expectations of its environment. We must verify theories and designs of sensorimotor systems empirically: the world keeps us honest and forces our assumptions to be reasonable. Such experimentation often yields insights and serendipitous discoveries. Integrated systems exhibit interaction effects that cannot be anticipated because they are not easily modeled. The cost of empirical testing must