Structured light detection and delineation of tripping hazards for people with impaired vision

Abstract

People with visual impairments (VI) experience more injurious falls and limited independent mobility than their normally sighted counterparts. Researchers have proposed several electronic travel aids (ETAs) that attempt to guide people with VI safely through unfamiliar environments, but these have all fallen into disuse. We highlight two key limitations in previous ETAs that contribute to this disuse- both limitations relate to ground plane tripping hazards. Previous ETAs have failed to 1) reliably sense ground level hazards and 2) transmit spatial information about the ground plane to the users in a manner that is useful for safe walking. We first designed a wearable ETA that can reliably detect tripping hazards and evaluated its performance on Boston city curbs. Second, we designed an augmented reality (AR) approach that allows a visually impaired person's residual vision to naturally behave like a structure light sensor, correctly obtaining salient 3D information about ground plane hazards. In a simulated environment, we found that curb recognition improved significantly, allowing subjects to identify curbs as small as one inch from a virtual distance of five feet. Finally, subjects with VI, wearing the prototype ETA, were evaluated on low contrast descending stairs on public routes in Boston. We found that with AR, their mobility was significantly improved based on walking speed metrics than without AR. As this ETA is hands free, does not require eye or head-ware, and allows people with VI to use their residual vision in a natural way, this approach could be accepted by people with VI, and therefor make a public health impact.--Author's abstract