Abstract
Humans walk to moving targets by turning onto a straight interception path that acheives a constant target-heading angle. Warren & Fajen (submitted) proposed a dynamical model of interception based on first-order information about target motion, which nulls change in the target-heading angle. The model successfully reproduces human paths to constant-velocity targets. Here we test model predictions for targets that change speed or direction. Participants walked in the VENLab, a 12m × 12m virtual environment with a head-mounted display (60 deg H × 40 deg V) and a sonic/inertial tracking system (latency 50 ms). Experiment 1 tests targets that accelerate or decelerate at a constant rate, on a trajectory perpendicular to the participant's initial heading. Ten within-subject conditions include two rates of acceleration, two rates of deceleration, two constant initial-velocity controls, and four constant mean-velocity controls. With an accelerating target, the model predicts a concave path that lags the target; with a decelerating target, it predicts a convex path that leads the target; alternatively, if participants make use of second-order information, they should anticipate the target motion. Experiment 2 tests targets that change speed or direction instantaneously in the midst of a trial, with an initial trajectory perpendicular to the participant's initial heading. Eleven within-subject conditions include two speed increases, two speed decreases, six direction changes (±45?, ±90?, ±135?), and a constant velocity control. The model predicts that each speed/direction change will be followed by a turn onto a new straight path that is specific to the target conditions.
NIH EY10923, NSF LIS IRI-9720327