Perceptual Invariants as a General Mechanism for Human Interception

Arizona State University, Scottsdale AZ

Investigators

Abstract

With National Science Foundation support, Drs. Michael McBeath and Thomas Sugar will conduct three years of psychophysical research with perception studies and robotic simulations to test and instantiate navigational principles that guide human interception of moving targets. They will use interception tasks to study intrinsic perceptual principles that use unconscious, invariant, viewer-based control variables. Interception tasks will be examined with targets above and below the horizon, in simple and complex hilly terrains, and with ballistic and non-ballistic trajectories. Three test domains will be included: (1) large-scale real-world tests with miniature head-mounted cameras to measure ongoing optical positions of target objects, (2) specialized motion laboratory tests replicated in a new state-of-the-art motion laboratory specifically designed to monitor body and projectile position using point light technology, and (3) robotic simulations to further establish the viability of the proposed perceptual models. The research will contribute to a unified fielder theory for interception based on control of simple perceptual invariants that can generalize across different types of interceptive tasks. The underlying invariant perceptual principles developed and modeled in this research will be applicable to a variety of other tasks such as the development of autonomous mobile robots, mobile tracking systems in automobiles and aircraft, and a better understanding of models of the pursuit of prey. Finally, the project includes synergistic research between perceptual psychology and robotics. The interdisciplinary research team with expertise in both perceptual modeling and robotics provides unique tools to succeed in this new and exciting approach to understanding perceptual principles. From this research, new principles will be discovered and enable a better understanding of generic navigational behavior that generalizes across many domains.