The idea is that the wheelchair would be able to understand voice commands and navigate to different areas, helping people with diseases such as multiple sclerosis maintain the highest possible quality of life and retain their independence, said Matthew Walter, a research scientist at MIT.
The project has been ongoing since 2007 as part of MIT's Computer Science and Artificial Intelligence Laboratory, which focuses on how robots can work in manufacturing, logistics, disaster relief, military applications and health care.
The project was spearheaded by an MIT student who realized that over time, people with neurological disorders lose their ability to walk and operate manual and joystick-operated wheelchairs, Walter said.
"What happens is they'll often run into walls, stationery objects, or people," said Walter, who has been working on the project for two years. "Do we continue as is and deal with the ramifications of collisions or eventually not allow the residents to operate their chair and instead rely on an available nurse to push them around."
Researchers are currently working on making the robotic wheelchair capable of understanding colloquial location terms, such as "kitchen" and "activity room" instead of traditional metric maps, which are like XY coordinates or blueprints.
"We've gotten fairly far," Walter said. He said the hope is that the wheelchair will be able to understand what it means when a person says "the gym is down the hallway," or "the cafeteria is next to the kitchen."
Much of the navigation and localization work is being done by Sachithra Hemachandra, MIT Ph.D. candidate, who has been working on the project for five years.
"For residents (in nursing homes) where their ability to get around independently is gradually eroding because of a disease, this has the potential to really change their life," Hemachandra said. "(But) there needs to be more effort to get it to the level that it can actually have that impact."
User studies are currently being done at MIT, and researchers there are working with residents at The Boston Home in Dorchester to see how the robotic wheelchair could benefit them.
Future plans include a user study at The Boston Home in about a year. That study would determine how people with no previous experience with the technology would describe a space to a robot, improving its voice-commandable capabilities. Aside from making the wheelchair understand voice commands, another challenge is making it look like a "normal" wheelchair without conspicuous computers and sensors attached to it, Walter said.
"We want to keep the amount of hardware that we put on it to a minimum," he said.
Researchers also added a collision-avoidance system and is working on adding mechanical arms to the chairs.
Whether the voice-commandable intelligent wheelchair will ever be commercially available for people with mobility issues is still uncertain.
"In the not-too-distant future, you can imagine having a small subset of users working with the product, but when it comes to a commercial product that's widely available, then it's really tough to say," Walter said.
That's because the technology is hard to scale, since people with progressive neurological disorders and mobility issues have varying needs when it comes to wheelchair specifics, he said.
Funding for the robotic wheelchair project has been supported over the years by Microsoft Corp., Taiwanese computer and electronic hardware manufacturer Quanta Computer and the US Army Research Laboratory.