Assessment of an augmented reality apparatus for the study of visually guided walking and obstacle crossing.

Abstract

To walk through the cluttered natural environment requires visually guided and anticipatory adjustments to gait in advance of upcoming obstacles. However, scientific investigation of visual contributions to obstacle crossing have historically been limited by the practical issues involved with the repeated presentation of multiple obstacles upon a ground plane. This study evaluates an approach in which the perception of a 3D obstacle is generated from 2D projection onto the ground plane with perspective correction based on the subject's motion-tracked head position. The perception of depth is further reinforced with the use of stereoscopic goggles. To evaluate the validity of this approach, behavior was compared between approaches to two types of obstacles in a blocked design: physical obstacles, and the augmented reality (AR) obstacles projected upon the ground plane. In addition, obstacle height, defined in units of leg length (LL), was varied on each trial (0.15, 0.25, 0.35 LL). Approaches to ended with collision on 0.8% of trials with physical obstacles per subject, and on 1.4% trials with AR obstacles. Collisions were signaled by auditory feedback. Linear changes in the height of both AR and physical obstacles produced linear changes in maximum step height, preserving a constant clearance magnitude across changes in obstacle height. However, for AR obstacles, approach speed was slower, the crossing step peaked higher above the obstacle, and there was greater clearance between the lead toe and the obstacle. These results suggest that subjects were more cautious when approaching and stepping over AR obstacles.