Is the U.S. facing an eldercare crunch? By 2050, the number of Americans 65 or older will balloon to 88 million — almost double the current tally — and it seems there aren't enough younger people to look after them all, especially as Alzheimer's disease and other cognitive impairments take their toll.
But maybe robots can take up the slack.
Thanks to advances in computer vision, natural language processing, and machine learning, as well as the falling cost of sensors, small and surprisingly capable robots are starting to show up in eldercare and nursing home facilities.
These humanoid bots can sit on a bedside table, providing reminders to take medication and even demonstrating exercise routines. But that's not all: they can answer simple questions like "What are we eating today?" and "What's today's weather?" Soon they'll be adding to their repertoire, and one day, they may be adaptable and inexpensive enough for home use.
"These robots are not there to replace humans," says Tommy Deblieck, CEO of Zora Bots, a Belgian firm that has pioneered the development of computer algorithms for eldercare bots. "They can't make any decisions. They carry out pre-set tasks or simple conversations to ease the strain on the nurses and make people's lives a little more pleasant."
Starting in December, Zora Bots will begin testing robots with the ability to stop people with dementia from wandering. These bots can read body language and determine whether the patient is confused. If so, the bots ask questions to distract the patient while sending a silent alarm to caregivers.
"In Belgium we have about 270,000 cases every year of dementia patients walking out of nursing homes and getting lost, resulting in a full-blown search mission with police helicopters, dogs, everything," Deblieck says. "We're hoping these robots will act as a second eye to prevent this happening."
Robot developers' ultimate aim is to create bots so smart and responsive that their elderly charges form an emotional attachment to them.
Research institutions around the world are working on this new breed of caregiving robots. One of the leaders is the University of Toronto, where Dr. Goldie Nejat, founder of the university's Autonomous Systems and Biomechatronics Laboratory, is working on a four-foot-tall wheeled robot that features a built-in iPad and a face that looks like a mannequin's, with orange eyebrows that light up during conversation. Called Casper, the robot is designed to work with institutionalized patients suffering from dementia.
Nejat is now fine-tuning Casper with the help of student actors pretending to be dementia patients. She watches as a test robot rolls over to a student, makes eye contact, and waves. "Hi, how are you doing today?" the robots asks in its synthesized voice. "Would you like to prepare lunch together?"
The robot then directs the student to the kitchen and asks if he or she needs help preparing the meal. If so, the bot can demonstrate via video displayed on its iPad. "That looks delicious," Casper says when the meal is ready — a prod that it's time to eat.
Between meals, Casper can facilitate phone calls with family members, engage patients in conversation, and organize games and other social activities. It can even provide clothing recommendations based on the weather.
"There's been some research which suggests that you can delay the onset or progression of dementia through repeated cognitive stimulation and social engagement," Nejat says. "So we thought, can you use a robot to increase the amount of stimulation these people get?"
How does Casper do it all? The robot features numerous sensors, including laser scanners that let it navigate its environment and avoid obstacles as well as 3D optical scanners that recognize everything from bingo cards to kitchen utensils. It also has microphones that can decode human speech even in noisy environments.
But what makes Casper really special is its ability to make complex decisions and to learn based on the reactions it elicits in the sometimes severely compromised humans with whom it interacts. The key is the robot's "brain," a microchip that's programmed to observe patients' speech, facial expressions, and body language and quantify their emotional states.
"To deal with these type of patients, it's really important that the robot is emotionally sensitive to some degree," Nejat says. "These patients can have good days and bad days so the robot has to be aware that now may not be the right time, and to try again on a different day."
Nejat hopes to make Casper and similar robots available to nursing homes and other eldercare facilities within the next few years at a cost of about $2,000. Her goal is to develop a robot that will work with family members to care for the infirm in their own homes.
"Allowing people to remain in their own homes and keep their independence would be a big step," Nejat says. "But at the moment, these robots are more social than physical. So they just prompt or encourage people to do an activity like making a meal, rather than actually doing it for them. But we're soon hoping to combine the social with the physical."
This would mean the robot needs grippers able to grasp and manipulate all sorts of household objects — and eventually capable of making beds, cleaning, cooking, and even supporting or lifting particularly frail individuals.
"It's a rapidly changing world," Nejat says. "Back in 2005, you couldn't buy a robot cheaper than half a million dollars, and it didn't have any intelligence either. It's an exciting time to be in the field."