By E.J. Mundell
Tuesday, November 18, 2008 12:00 AM
TUESDAY, Nov. 18 (HealthDay News) -- In the not-so-distant future, American seniors may turn to helpful, uncomplaining robots to fill the worrisome "care gap" that many face today.
One of these autonomous devices, called the uBOT-5, is already capable of carrying out simple tasks while it monitors the home environment. The robot can even spot trouble -- such as a person falling down -- and call 911 if necessary.
The freestanding device can also bring a faraway loved one into an aging person's home via video Internet hook-up.
"So, if I'm at work, and it's lunch hour and I want to poke in on Dad, I can get on the Internet and basically 'step inside' the robot," said uBOT-5 co-inventor Rod Grupen, who directs the Laboratory for Perpetual Robotics at the University of Massachusetts, Amherst. With their face appearing via video on the front of the robot's head, the virtual visitor can converse with their loved one while moving the robot around, doing some cleaning, for example, or retrieving a dropped TV remote.
Any "authorized user" can jump into and guide the robot, Grupen said. "So, if you can't get to your doctor, your doctor can now come to you," he said. In fact, the UMass team hopes that the uBOT-5 will someday be capable of running simple medical tests, such as measuring blood pressure or blood sugar.
And because it's fully mobile, with Segway-like wheels, virtual visits from others should include much of the house, and beyond. "Your granddaughter on the West Coast can get into the robot and visit with you outside in the garden, you can have a two-way conversation with audio/video, hold hands and go show them the flowers you just planted," Grupen said.
There's a huge and growing need for robotic home assistants that might help care for the elderly or disabled and allow them to stay in their homes, Grupen believes. According to U.S. Census figures, the number of Americans age 65 or over will double by 2030, and two-thirds will need some form of long-term care. At the same time, there's a dearth of nurses and home health-care aides to care for them; experts predict a shortage of 800,000 nurses by 2020.
The uBOT-5's design was inspired by the human body. Its myriad sensors mimic human eyes and ears, constantly scanning its environment. It is even programmed to detect and respond to worrisome aberrations, including a fallen, unresponsive human. The robot's arms are each capable of handling 2.2-pound loads, and they can extend to reach high or pick things up off the floor (a dropped pill bottle, a package in a foyer, for example). The robot can lie prone to scoot itself under a bed (and then right itself), and it may even someday help with household cleaning and grocery shopping, Grupen said.
And the cost? Right now, the prototypes at UMass cost $65,000 apiece, but Grupen envisions a day when commercial versions would be sold for $5,000 plus a monthly Internet hook-up fee, much like today's computers.
And the uBOT-5 isn't the only such device in the pipeline. Over at Massachusetts Institute of Technology, researcher Nicolas Roy, at the institute's Computer Science and Artificial Intelligence Laboratory, has developed an "autonomous wheelchair" that only requires a command to whiz users from one spot to another in a hospital or nursing home.
When first delivered to a facility, the wheelchair -- rigged out with high-tech scanning software -- has no knowledge of the particular layout. But staff will uncrate it, turn it on, and give it a verbal guided tour, walking it past different rooms and nursing stations.
"You talk to it like you'd talk to a new person, a new nurse. And as a side effect of the thing being walked through the facility once or twice, the wheelchair has now been demonstrated a route between all the points," explained co-developer Seth Teller, who helps lead the lab's Robotics, Vision and Sensor Networks Group.
After that, a wheelchair-bound stroke patient or quadriplegic need only say, "Take me to Room 451" for the chair to understand and then do just that. The device will be launched as a prototype ready for testing in a Boston-area nursing home within two years, Teller said.
Finally, at Georgia Tech, researchers led by assistant professor Charlie Kemp are making their own home-care robots, inspired by the agile intelligence of service dogs.
"We're using service dogs to answer three important questions: What tasks would be good for a [home] robot to perform? How should people interact with the robot, to tell it to do these tasks? And how can the robot actually perform these tasks, given the complexities of the home?" Kemp said.
Service dogs and the disabled people they help are providing the answers. The new robot is being designed to move about and perform tasks such as opening drawers, turning doorknobs and working light switches, Kemp said. Users indicate what they'd like done by using a laser pointer, and homes are modified slightly to help the robot, just as homes are subtly tweaked to aid service dogs. "Things like tying a small towel to a doorknob" to facilitate grasping, Kemp explained.
The robot may not ever replace a great service dog, but Kemp noted that the average disabled American now pays $16,000 for a properly trained canine, and waiting lists now stretch for years.
"I think there's a real need," he said. "So, the hope is that people will support this sort of work. Then, we'll be able to deliver these things when people need them."
There's more on the uBOT-5 at the University of Massachusetts, Amherst.
SOURCES: Rod Grupen, Ph.D., professor, computer science, and director, Laboratory for Perceptual Robotics, University of Massachusetts, Amherst; Seth Teller, Ph.D., professor, co-head, Robotics, Vision and Sensor Networks Group, Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge; Charlie Kemp, Ph.D., assistant professor, department of biomedical engineering, Georgia Institute of Technology, Atlanta