Wearable computer vision aid helps blind people avoid collisions

A team of researchers showed that people using a long cane or a guide dog were able to reduce the number of collisions by 37 per cent, compared to using other mobility aids alone.

“Independent travel is an essential part of daily life for many people who are visually impaired, but they face a greater risk of bumping into obstacles when they walk on their own,” said Gang Luo, an associate professor of ophthalmology at Harvard Medical School.

“Although many blind individuals use long canes to detect obstacles, collision risks are not completely eliminated. We sought to develop and test a device that can augment these everyday mobility aids, further improve their safety.”

The experimental device and data recording unit were enclosed in a sling backpack with a chest-mounted, wide-angle camera on the strap, and two Bluetooth-connected wristbands worn by the user.

The camera is connected to a processing unit that captures images and analyses collision risk based on the relative movement of incoming and surrounding objects in the camera’s field of view. If an imminent collision is detected on the left or right side, the corresponding wristband will vibrate; a head-on collision will cause both wristbands to vibrate.

Unlike other devices that simply warn of nearby objects whether or not a user is moving towards them, this device analyses relative motion, warning only of approaching obstacles that pose a collision risk, and ignoring objects not on a collision course.

The study involved 31 blind and visually impaired adults who use either a long cane or guide dog (or both) to aid their daily mobility.

After being trained to use the device, they used it for about a month at home in conjunction with their typical mobility device (mostly a long cane).

The device was randomised to switch between active mode, in which the users could receive vibrating alerts for imminent collisions, and silent mode, in which the device still processed and recorded images, but did not give users any warning even if potential collisions were detected.

The silent mode is equivalent to the placebo condition in many clinical trials testing drugs. The wearers and researchers would not know when the device modes changed during the testing and analysis.

Collisions were analysed by researchers from the recorded videos and the effectiveness of the device was evaluated by comparing collision incidents that occurred during active and silent modes. The study found that the collision frequency in active mode was 37 per cent less than that in silent mode.

Alex Bowers, a clinical researcher and one of the co-authors of the paper, said the video recording from the study also provides rich data about daily life mobility of people with visual impairments.

“Long canes are still very helpful and cost-effective tools that work well in many situations, but we hope a wearable device like this can fill in the gaps that the cane might miss, providing a more affordable, easier to obtain option than a guide dog,” he said.

The team are now planning to make the device smaller and more cosmetically appealing.