The makers of the Argus II retinal prosthetic have devised a way to help blind people read Braille. Their first reported tests with a person wearing an Argus II indicate that this method could help blind people with the implant read signs and short sections of text.
Second Sight’s prosthetic are built for patients with degenerative eye diseases like retinitis pigmentosa. The Argus II eye prosthetic captures images through a camera positioned on the wearer’s face. It is connected to an electrode array implanted on the eye. Images seen by the camera are converted to electrical signals and transmitted to the electrode array, which stimulates nerves in sections of the retina.
This gives the wearer a real-time replay of the world in front of them. But though they see better with the prosthetic, the view they have is a fuzzy one, which makes text and street signs particularly hard to read.
Second Sight’s solution is to stimulate electrodes in the pattern of Braille letters instead of standard text. This technique would replace the video processing section of the Argus II with a kind of Braille visualizer, to convert alphabets into their corresponding Braille arrangement of dots. If a wearer is facing a signboard with text on it, it’s the Braille lettering they would see.
A Second Sight team led by Thomas Lauritzen reports in the Frontiers in Neuroprosthetics that such an approach shows promise as a viable text reader. In their first tests of the system, the team activated a section of the electrode array in a patient implanted with the Argus II, stimulating nerves to give a visual perception of “seeing” Braille letters.
The test subject, who is blind from retinitis pigmentosa, was able to recognize single letters 89 percent of the time, and short words as well: he correctly recognizing 80 percent of 2-letter words, 60 percent of 3-letter words, and 70 percent of 4-letter words (As the Second Sight researcher told New Scientist, software that converts letters to Braille already does exist.)
Recognizing the Braille letters for now does take longer than reading the words with fingers from a page. But it’s fast enough to read signs and short messages, the team expects, and expects to reach speeds of about 120 letters a minute.