Preliminary evaluation of a wearable video camera based collision warning device for blind individuals in an obstacle course

Abstract

Purpose : A wearable video camera based collision warning device providing audio warnings was previously shown to help people with peripheral vision loss avoid collisions in an obstacle course. A new version of that device was developed for people with severe vision loss or total blindness providing collision warnings via two vibro-tactile wristbands. We evaluated this new device in an obstacle course to determine its benefit in avoiding collisions.

Methods : A 45 m long indoor obstacle course was set up in the form of an ‘L’ shaped loop with 12 overhanging obstacles (> 1 m above the ground). Twelve subjects, 8 blindfolded normally sighted (NS) and 4 blind, walked with a long cane following the boundary of the obstacle course in two conditions: with (WD) and without device (WOD). In WD condition, the device was mounted on the chest and the task was to avoid collisions based on the warnings received from the device. Subjects walked the course in opposite directions WD and WOD. The device order was counterbalanced. Training was given prior to the task. NS were given additional cane usage training. Number of head-on collisions and task completion time in each condition were recorded by the experimenter. Video from the device camera with collision risk localization information was logged. Baseline walking speed was measured WOD along an unobstructed corridor and percent preferred walking speed (PPWS) was computed.

Results : Median values [25^th – 75^th percentiles] are reported for all quantities. The number of collisions reduced significantly with the device (WD: 4 [3 – 4.5], WOD: 11.5 [10.75 – 12]; p < 0.001). Except for 1 blind subject who had the same number of collisions in both conditions, all subjects experienced fewer collisions WD. Video recorded from the device showed that it failed to identify a median of 2 [2 – 3.5] collisions, which is lower than the observed collisions WD. This indicates that subjects might not have reacted to some of the collision warnings. There was a small but significant reduction in PPWS with the device (WD: 36 [31 – 41], WOD: 40.9 [39 – 45]; p = 0.04), possibly because the subjects attempted to navigate around the obstacles.

Conclusions : The results of this pilot evaluation are promising. The number of collisions reduced by 65% overall with the device in the indoor obstacle course setup.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.