We've written about emerging technology for visually impaired people before, in the form of a new, more versatile kind of "refreshable braille" display that could one day provide tangible pictures and graphs to its users. You can read that article here. Today we have a new piece of assistive technology to feature: OrCam, an already-available device that can help visually-impaired users by identifying nearby objects and even reading text.
The device works by taking a photo with a small camera that attaches to a user's glasses, then describing objects in the photo to the user through a small bone-conduction speaker which also sits atop the user's glasses. This combination camera-speaker part of the unit is connected by a wire to a larger rectangular box, which can be placed in a pocket or bag. Unlike other solutions designed for similar purposes, OrCam is completely external and does not require any implants.
So how does OrCam know when to take a photo? It's user-activated, meaning that the person wearing it has to make a special gesture (in this case, a pointing sign) within the camera's field of view in order to activate the software's special features. Once the photo has been taken, OrCam will analyze and attempt to identify whatever the user was pointing at. If the object contains text (a street sign, for example, or a label or even an entire newspaper), OrCam can even read that text aloud. Watch this video from CNN Money to see it in action:
The man demonstrating how the device works in the video above is Amnon Shashua, who cofounded OrCam in 2010 after being inspired by his wife's aunt. He shares the story in this TED@NYC talk, saying that his aunt-in-law had macular degeneration, a condition which causes degradation of the retina and loss of vision. Shashua says he came up with the idea for OrCam when she talked to him just after his doctoral ceremony at MIT. A computer scientist specializing in the artificial intelligence field, Shashua initially didn't take her request for help seriously, saying that though he was a doctor now, he wasn't the kind who helped people. But the idea never truly left his mind, and a few years later, OrCam was born.
Elements of Shashua's artificial-intelligence background are still very apparent in some of OrCam's features—the software is not limited to identifying only those objects which it has already been programed to identify, but can also "learn" to understand new objects with some simple instructions from the user. Given time, it can be customized to suit a user's particular environment, even identifying faces (OrCam has made this video demonstrating those features).
It's not a perfect system yet; the device has trouble in low or bright lighting, cannot read handwriting or script fonts, and currently only "speaks" in English. But Shashua and his team are continuing to update and improve OrCam, and for many visually impaired individuals, it's already life-changing.
You can learn more about OrCam in The New York Times, ZDNet, Bloomberg News, and other places as listed on their website.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment