William H. Warren
Dept. of Cognitive & Linguistic Sciences
Brown University
Mauldin Auditorium (NSH 1305)
Refreshments 3:15 pm
Talk 3:30 pm
Humans, animals, robots, and animators all face a similar problem: how
to generate paths of locomotion through complex changing environments.
I argue that locomotor paths are not explicitly planned, but emerge from
local interactions between an agent and a structured environment. I
will describe an empirical model of steering and obstacle avoidance that
accounts for locomotor path formation in humans.
Locomotor behavior is composed of four elementary components: (a)
steering toward a stationary goal, (b) avoiding a stationary obstacle,
(c) intercepting a moving target, and (d) avoiding a moving obstacle.
Based on experiments in an immersive virtual environment, we developed a
dynamical model of each component that generates realistic locomotor
paths. Goals behave as attractors of the current direction of heading,
whereas obstacles behave as repellers; moving targets are intercepted
by nulling the change in the target-heading angle, and moving obstacles
are avoided by treating this null point as a repeller. These basic
components can be linearly combined to predict human routes in more
complex environments.
The results demonstrate that locomotor path formation can be understood
as emergent behavior that arises from interactions between a structured
environment and an inertial agent with a few simple control laws, making
explicit path planning unnecessary. The model may also be used to
generate biologically realistic locomotor paths.
Bill Warren is Chair of the Department of Cognitive and Linguistic
Sciences at Brown University. He is also co-director of the Virtual
Environment Navigation Lab, the largest immersive VE in the known
universe for the next 15 min. His research focuses on the perceptual
control of human action, including optic flow and the control of
locomotion, path integration, and visual navigation, as well as such
feats as bouncing a ball on a racquet, catching fly balls, and infant
bouncing. He is the recipient of a Fulbright research fellowship, an
NIH Research Career Development Award, and Brown's teaching award in the
Life Sciences.
Please contact Ruth Gaus (riw@ri.cmu.edu) for appointments.
The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.