Abstract
When moving through the world, humans often encounter moving obstacles that threaten to impede their path. Fajen & Warren (2004, in press; Cohen & Warren, VSS 2005) proposed a dynamical model of moving target interception and moving obstacle avoidance based on first-order information about object motion, requiring no higher-level path planning. This model has been supported in instances of novel target and obstacle motion; however, repeated exposure to one target trajectory (but not two) can lead to anticipation of the target's motion and a more direct interception path (Owens & Warren, VSS 2005, 2006). Here we investigate whether people can learn to anticipate more than one moving obstacle trajectory, whether this is aided by obstacle shape and color cues, and whether explicit instruction on these cues can aid learning. Participants walk 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 50–70 ms). On each trial, participants walk to a stationary target at 6m while avoiding an obstacle that moves across their path. Participants complete three blocks of trials - one control block with constant obstacle speed and direction, one block in which obstacles change direction mid-trial, and one block in which obstacles change speed mid-trial. Each obstacle trajectory is cued by its shape and color. The Number of obstacles per block (2 vs. 3) and Instructions (cue-explicit or neutral) are varied between subjects. Preliminary data indicate that participants can learn to anticipate two obstacle trajectories without explicit cue instructions; it is expected that the addition of cue instructions will increase the speed of learning and the number of trajectories learned. The results suggest that people can learn to anticipate the motion of more than one obstacle if given sufficient information.