Out of the Darkness

Imagine walking into a crowded party. You want to find a place to sit down or perhaps find a familiar face and start up a conversation.

Now imagine that in front of you is nothing but darkness. Your ears are overwhelmed with the sounds in the room—people talking, shuffling feet, glasses clinking, chairs scraping across the floor. The problem is that you have no point of reference.

You know that there are people around, but you have a distinct feeling of isolation. You are unable to scan your surroundings or even distinguish between the front or back of the room.

Such a scenario is unlikely for most people. But it is a daily reality for the visually impaired. Sethuraman Panchanathan and his colleagues at Arizona State University are working on a series of innovations that could brighten that darkness.

Panchanathan is a professor of computer science and engineering. At ASU he is known as “Panch.” In 2000, he helped launch ASU’s Center for Ubiquitous Computing (CUbiC). The goal is to develop new technologies that allow implicit interactions with computers. By implicit, Panchanathan refers to developments such as a smart house or a smart car—technology that allows a device to sense a person’s needs without a vocal or keyboard command.

As Panchanathan and ASU colleague Kasim Candan developed the research agenda, they realized something important was missing. They decided they needed a human face to go with their science.

“I’ve always felt that if my technology is not useful to people, then it’s not really worth doing,” Panchanathan says.

Candan suggested developing devices to aid the visually impaired. The instant the suggestion was made, there was agreement. The researchers had found their inspiration.

“For years I’ve worked with processing images for those who could see. Never for a moment had I thought about providing images for those who could not see,” Panchanathan says. “The challenge was right in front of us. I knew this was a golden opportunity to make a difference.”

The ASU researchers moved quickly to identify ways in which to help the visually impaired. At a chance off-campus meeting, Panchanathan ran into Terri Hedgepeth, a student support specialist with ASU Disability Resources for Students (DRS). Hedgepeth instantly became a planning resource.

Hedgepeth is visually impaired. She helped the computer scientists set up several focus groups. She wanted people who would use the technology to explain exactly what they really needed.

“What we heard were needs that none of us even thought would be important,” Panchanathan says.

Initially, the researchers thought navigation would be the key issue. But results from the focus group identified a specific need. They learned that a device was needed to help the visually impaired in basic social situations as well as in breaking the barriers to the printed page.

Without sight, a person does not know who is approaching. In a crowded party room, it might be difficult to impossible to quickly assess whether friends are present. One focus group member had lost his sight as an adult. He suggested a device that could help him identify attractive women in a bar.

“Of all the things he could want, I thought,” Panchanathan says. “But why not? Then I got it. This project should provide a means to help the visually impaired experience life as a normal human. That is the challenge that drives us.”

The ASU team’s work-in-progress includes the development of a group of iCare devices. The iCare-HumanRecognizer includes a small computer processor linked to a camera mounted on either glasses or a headband. The camera takes a picture of a person and quickly compares it to other images stored in database to identify the person. A message is then sent to the user via audio output or a refreshable Braille keypad.

If the person’s image is not in the database, the device might describe the person and add them for future reference. Using the iCare-HumanRecognizer,, a visually impaired user could identify an approaching friend to say hello, or simply initiate contact with someone new.

The device could help a student to quickly find an empty seat in a crowded lecture hall. There are personal safety implications as well. The iCare-HumanRecognizer would allow the user to assess the physical description and body language of an approaching stranger.

Other members of the focus group suggested the need for a device that would allow the visually impaired to read books not available in Braille. Many books have not been translated, and listening to audio books can be time consuming. Another person mentioned the lack of ability to quickly scan newspaper headlines. Another talked about her inability to compare nutritional values on food items while shopping for groceries. A handheld device that a person could point at a book, or any block of text, to read the text would remove those barriers.

The researchers listened. They also are developing a second device called the iCare-Reader. The full-size tabletop device will quickly scan a page of text and translate into either audio or Braille. The device would also allow the user to enter the text at any point—mid-book, mid-chapter, mid-page, or even mid-sentence—which is another limitation of audio tapes. Once perfected, the device would be reduced to a handheld size.

Panchanathan’s team includes senior scientists and graduate students. They have dedicated themselves to reaching specific benchmarks every few months. The ASU scientist says that he wants accountability for their work.

The challenges are great, the obstacles many. Can the devices be made to work in different lighting scenarios? Can the devices overcome poor text quality? How do they make sure a book being scanned is right side up? How do they create an audio output that does not interfere with the person’s ability to hear what is going on around them?

“We wanted to do something difficult, something that would be a real challenge,” he says. “It is the difficult things that are worth doing.”—Gary Campbell