Abstract
How do we perceive the behavioral intentions of another pedestrian? In this study we investigate whether participants can perceive the intention to pursue or evade based on the trajectory and humanoid appearance of a virtual avatar. The steering dynamics model of target interception and avoidance does a good job of describing pursuit and evasion of pedestrians (Cohen, Cinelli, & Warren, VSS 2007). Here we use the model to drive an avatar so it pursues or evades the participant.
Thirteen participants wore a head-mounted display (63° H × 53° V) and walked in a virtual environment towards a goal while an avatar approached. The task was to evade avatars that appeared to be pursuing them, or pursue avatars that appeared to be evading them. We manipulated the avatar's trajectory (evaders moved toward a goal 2, 4, or 8° from the participant's initial heading and appearance (textured post or walking human). Head position was recorded using an inertial/ultrasonic tracking system (IS-900, 70 ms latency).
The percentage of trials in which evasion was detected increased with the avatar's approach angle (F(3, 10) = 19.04, p F(1, 12) = 7.33, p F(3, 10) = 2.92, p p p = 0.005), perhaps because its head and limbs were aligned with the direction of travel (heading). Future work will investigate the contribution of such heading cues, head turns, and the contingency of the avatar's trajectory on the participant's movements.
Supported by NIH EY10923.