Speaker: Brent Fajen, Rensselaer Polytechnic Institute.
Title: Optic flow-based perception and control in complex, dynamic environments
Abstract: Humans and many other animals are capable of moving, oftentimes at high speeds, through complex, dynamic environments based primarily on visual information. Research on visual control has revealed optic flow-based strategies for perceiving and controlling self-motion and informed the development of computational neural models of optic flow processing. Such work not only advances our understanding of sensorimotor control in biological systems but also has the potential to inspire new solutions for vision-based control in aerial robots. In this talk, I will explore some of the promises and challenges of research at the intersection of these two areas. The focus will be on solutions that better capture the robustness and flexibility of optic flow-based perception and control in humans and have the potentialto be adapted for use on board small aerial robots. I will present results from three studies. The first considers the hypothesis that the robustness and stability of human heading perception is rooted in sensitivity to temporally evolving optic flow rather than a snapshot of the instantaneous flow field. The second study develops a self-tuning mechanism for maintaining sensitivity to information in optic flow across wide variations in conditions. The third project, which is a work-in-progress, investigates how drone-racing pilots use active gaze to efficiently sample information in optic flow while steering at high speeds through cluttered environments.
Biography: Brett Fajen has a B.S. in Mathematics from Bucknell University (1993) and a Ph.D. inExperimental Psychology from the University of Connecticut (1999). From 1999 to 2001, he did a post-doc with Bill Warren at Brown University. In 2001, Brett joined the Cognitive Science Department at Rensselaer Polytechnic Institute in Troy NY, where he is currently a Professor. Brett’s research investigates sensorimotor control, with a focus on the visual control of locomotion in complex and dynamic environments. He studies a variety of problems within this area, including the perception of affordances,self-motion, and object motion, and the control of steering, interception, obstacle avoidance, and foot placement during walking over complex terrain.