Logo: University of Southern California

The New Seeing-Eye Dog

Although millions of Americans cannot see, USC Viterbi Professor Gerard Medioni’s new “seeing-eye” technology can help them get around safely, making their lives easier.
By: Anna-Catherine Brigida
October 01, 2013 —
Medioni demonstrates how to use his technology. 

From the moment we open our eyes in the morning, our brain processes, seemingly effortlessly, what we see. Eugenia Barrera does not have this luxury. For her and the estimated 6.5 million blind or visually impaired Americans, life is more complicated.

Even in her own home, Barrera struggles to navigate and find specific items. Stepping into an unfamiliar environment can elevate these difficulties from somewhat challenging to incredibly stressful.

USC Viterbi Computer Science Professor Gerard Medioni wants to make it easier for her and other visually impaired people to get around.

Computer vision specialist Medioni and Biomedical Engineering Professor James Weiland, along with their teams, have created a pair of glasses and a vest that do just that.

This glasses-vest combination is a type of “seeing eye” for the blind. The glasses observe the environment and the vest gives directions, much like a seeing-eye dog sees and leads.

The glasses create a 3D image of the surrounding environment using two cameras. Although the blind person cannot see this image, the computer can, using it to analyze the setting.

With this image, the computer can distinguish between obstacles and open space. Using this information, the computer sends signals to a vest through four sensors, much like the brain sends messages to the body.

Getting close to an object triggers a warning buzz. If a person is too close to a table on their left, the lower left sensor would buzz, telling the person to move away from that object. No sound means all clear.

Medioni and Weiland’s team started out in 2008 with the simple task of autonomously navigating an obstacle course, originally funded by an NSF grant. The project is currently funded by the Telemedicine and Advanced Technology Research Center. Through collaboration with the Braille Institute, Barrera tested this technology in a parking lot. It gave her a sense of independence she had not previously experienced.

"For me, it was very freeing," she said. Iit gave me freedom to not have to ask so many people for help. I can be more independent."
The project was largely successful, but Medioni and his team did not stop there. They realized that life is not a static obstacle course made in a laboratory.

In real-life scenarios, obstacles are not always stationary. Often, they are moving people. Still, people will move out of the way. So, Medioni and his team advanced their technology to differentiate between threatening obstacles-buildings, branches, or benches- and non-threatening obstacles such as people.

To test this technology, Medioni and his team imagined the most hectic scene possible: USC campus on a football game day, when swarms of people can overwhelm even someone with 20/20 vision. The glasses and vest successfully led a team member to a destination without the vest buzzing incessantly.

The next real-life obstacle course the team will test is a grocery store. The glasses-vest combination will lead the test subject to a specific item in the grocery store for them to pick up. The test will take place in October.

Medioni enjoy his research and how it benefits people in need.

“The best part of it all is that people actually like the system,” Medioni said. “We really started from the beginning to involve patients and get their feedback and make sure that this was something that would be helpful and useful.”

Based on this feedback, Medioni and his team added one more layer of complexity to the problem: communication. They recognized that lack of vision makes it difficult, if not impossible, to understand non-verbal communication such as hand gestures or facial expressions. Medioni wants to change that by addressing the most complex and nuanced issue yet.

This would allow a blind person to fully understand the situation, such as a meeting at work. With facial recognition technology, they would be able to tell who is being addressed based on where someone is looking or how serious or joking the speaker is based on facial expressions. Medioni is experimenting with ways to communicate this information to the blind person, potentially using the vest again or through a speaker in their ear.

“They want to be part of the working environment,” Medioni said. “They want to be inserted in society and that’s really what we are trying to help them accomplish.”

The project is still in its early stages.

Barrera for one has become a big fan of Medioni’s work and its implications for people like her.

“I keep telling them to hurry up and finish,” Barrera said. “I need this in my life.”