Meet your robot caretaker

Roberta does all this with ease, like the well-oiled machine she is. You’d expect nothing less from your robot caretaker.

Of course, a person could also do all of this work, and right now, people do. But humans are expensive, even as the job of caring for older people pays poorly. The 2 million home healthcare aides in the United States — 90% are women and over 50% are people of color — make an average of $12 an hour, a quarter of them live below the federal poverty line, and half rely on public assistance to get by.

Not only are home healthcare aides notoriously low-paid, they perform a dangerous job: Lifting and moving people’s bodies to wheelchairs, toilets, and baths is part of the physical work required and results in a “high risk” for injuries, according to the Centers for Disease Control and Prevention. The intimate nature of the work and needle sticks can lead to disease transmission from elderly patients. It’s a tough job with high turnover.

At the same time, the elderly population is expected to double by 2050. “We already have a shortage of home health nurses and aides, and life expectancies keep getting longer,” said Dr. Katherine Pannel, a geriatric psychiatrist and the medical director of Senior Life Solutions in Winona, Mississippi. Some kind of in-home care is necessary to keep costs down — the 14 million Americans who need assistance but live on their own save billions for Medicare and their own pockets compared to those who live in nursing homes. “So, I think robots and artificial intelligence certainly has a role in the future care of geriatric populations,” said Pannel.

You might blanch at the idea of replacing people with robots, but human home healthcare aides have their limitations. Consider that the connection between a patient and his or her carer, like any human relationship, can be fraught for any number of reasons, including plain old personality conflicts. There’s also the fact that patients will care about the caregiver, which sounds nice but can also mean that reticent patients don’t ask for what they need, which can be stressful. In a survey of elderly Japanese, one of the reasons cited for preferring a robot carer was that “not having to worry about a robot’s welfare was a plus.”

And, of course, “to err is human.” People make mistakes, get sick, have bad days, or have to care for their own children or parents and can’t make it into work sometimes, leaving the vulnerable elderly to fend for themselves. And even worse, elder abuse is not uncommon (between 7%-10% of the elderly are abused by family or carers).

If and when this change takes place, there are reasons to view it not as part of a sci-fi dystopia but as an improvement in societal well-being.

Unlike humans, robots can be programmed to be consistent and reliable. They can respond at any time of day or night, while human caretakers are usually on duty only during specific times. While their charges sleep, a robot can do household chores. Robots won’t forget to help with exercises, dispense medicine, or do needed household tasks on time, reliably.

If robots make a mistake or do something that causes pain, it’s simple to track and fix, without the guilt and shame that causes humans to cover problems up. And they don’t hold grudges or take out feelings on their charges. Even the best of humans can get frustrated and angry with an older person who may be neurologically challenged or in pain and lashing out. A robot won’t have feelings to hurt.

Then there is the very intimate work that can be awkward for both. Robots don’t get embarrassed, and honestly, who would you rather have give you a bath — a person you barely know or a robot that has been programmed to know how much soap you like?

“Robots could do things like check vital signs: blood pressure, blood sugar, and pulse. Patients get confused by when to take medications, and that can be deadly — and it makes them anxious. A lot of home health aid work, that is all they are doing — filling pillboxes, drawing up insulin, and household tasks,” said Pannel.

She says robots handling these tasks would free humans to do other, more important work, the kind only humans can do: “How awesome would it be for the home healthcare nurse to play games, discuss TV shows, take them outside for fresh air, take them to get their hair done, instead of mundane tasks?”

This kind of human connection is something that the isolated elderly, who don’t have family or social support, desperately need. “Some of my patients only have social interaction during doctor’s appointments, so increasing that human interaction time is really important,” said Pannel. “We don’t want robots replacing that.”

For those elderly people who do live with a partner or other family, a robot could similarly free up time and reduce stress, becoming a part of the family that allows the “sandwich generation” of working parents who have both children and elderly parents at home more time to engage in enriching, enjoyable activities with children and grandparents.

In some ways, the future is now. The first generation of consumer-facing, caretaking robots is already here, or is being tested. There’s Mobiserv in Europe; GiraffPlus in Sweden; Stevie, designed in Ireland; and Zora in France. In Japan, where 80% of the population has a positive view of robots for elder care, a government-backed project assisted “98 manufacturers to test nursing-care robotic devices … 15 of which have been developed into commercial products.” And of course, Paro, the plush comfort robot “seal,” is already popular in Japan, as well as North America and Europe.

And there’s Temi, a robot that was going to be rolled out as part of a new luxury senior living center in Manhattan. When the coronavirus hit, these robots were immediately utilized at the company’s suburban locations, according to Shane Herlet, CEO of Maplewood Senior Living. The company deployed the Temis to video chat with families, doctors, and the internal teams, deliver packages, and pick up items left behind in the dining room. Temis are even learning jokes. “While the robots … normalize daily life with strict social distancing and social isolation rules, the real benefit that we were hoping for and are seeing is their ability to lift the spirits of the residents. We see our residents dance to the music Temi plays, laugh along with its jokes, and utilize the video chat function to see family members that are temporarily not able to visit,” said Herlet.

These robots aren’t doing chores yet; all of the above devices are, in some way, social. Paro and Zora are mostly pet replacements that, due to their high costs, are usually shared by a nursing home community. Paro is especially calming to dementia patients, effectively replacing anxiety medications in some cases. Unlike a real animal, Paro can’t be hurt by a confused patient.

GiraffPlus, Mobiserv, and Stevie all combine some social aspects but are aimed more at reminding and motivating their humans — to eat a snack, drink water, take medications at a certain time. Only one robot prototype can lift and move a human body so far: the RIBA from Japan’s RIKEN-TRI Collaboration Center for Human-Interactive Robot Research. And none cook or do chores.

While it’s still early for more widespread robot-human adoption, studies show that for most people, connections to artificial intelligence or even mechanical objects can happen without even trying (as anyone who has named a car and cried when it was sold will attest). Dr. Elizabeth Broadbent, professor of health psychology at the University of Auckland, has studied human-robot interaction for over 20 years and found that most people became friends with their healthcare robots. “This has happened automatically as people have liked having the robot in the house and found it provided a kind of social presence through talking and moving and blinking lights,” she said.

How much a patient needs to “bond” with his or her robot depends on what it’s needed for, says Tatsuya Nomura, professor in the Faculty of Advanced Science and Technology at Ryukoku University in Japan. He studied these interactions and found that for purely physical tasks such as eating and going to the bathroom, a more mechanical robot that a patient doesn’t have an emotional connection to can be best since that connection can lead to “care receivers’ avoidance of the robots” out of embarrassment. But when it comes to social tasks, such as aiding the “management of medicines, emotional connection with robots may encourage the task,” said Nomura.

Integrating a variety of physically capable robotics with artificial intelligence that is more complex than basic companionship or a nagging machine will take time. “We will need about 50 years to develop that,” said Dr. Hirohisa Hirukawa, director of the Robot Innovation Research Center at Japan’s National Institute of Advanced Industrial Science and Technology.

Trying to integrate the bulk strength needed to lift another person, intermediate power to pick up a set of sheets from the stack in the linen closet, and delicate precision to comb hair or count out pills is a tall order for a machine. And safety is a serious issue.

A more powerful robot can do more kinds of jobs, so for a long time, the thinking was the stronger, the better. Not so fast, says Chris Atkeson, a professor of robotics in the Human-Computer Interaction Institute at Carnegie Mellon University. The concern isn’t that a robot will suddenly go berserk and act out against humans in some kind of sci-fi nightmare. It’s about patient care: “You don’t want to use a bulldozer to brush your teeth. As you make a robot stronger, they are going to be more and more like a bulldozer — it can lift heavy loads, but it doesn’t have a delicate touch. That’s the laws of physics. You’re either weak with a delicate touch or strong and kind of dangerous.”

After all, “These are fragile people, and we want to be gentle.” Working these challenges out is what puts a fully humanoid robot that can do it all on the long-term plan.

One question on everyone’s mind: What will the robots look like? Broadbent says to expect form to follow function: “You expect a doctor or nurse to tell you to take medications, so a robot that reminds you to take medications needs to look and act like a doctor or nurse. We know that cats can’t talk, and they certainly know nothing about medications, so people won’t take medications from a talking robotic cat. However, if we want a pet for companionship, then we want a robot to look and act like a cute cat or other animal.”

Experts who study the ethics of robotics have found that human-like robots aren’t ideal — they’re much more likely to be rejected than a robot that seems like an animal or just looks like something entirely new. That’s down to the “uncanny valley” aspect of robotics, in which a close-but-not-quite human creeps people out.

“Something like C-3PO from Star Wars has a human-like form but is clearly not human due to his metallic surface. Getting the right mixture of function and form that makes us feel comfortable is definitely important,” said Broadbent.

All but the most basic robots used in healthcare are already equipped with artificial intelligence that can, over time, “learn” the humans they work for. This will lead to specialization far beyond the physical presentation of the robot. An isolated person who craves company could have a machine that chats or reads. Patients could request funny, intellectual, or quiet robot personalities. Some might focus on keeping a person physically active, while others might challenge their person with games and puzzles to keep cognitive skills up. Personalization will be key to providing good care and building meaningful psychological relationships between people and robots.

As robotics technology and AI advance over the coming decades, researchers say that before we have a fully functional robot caretaker, we’ll see a variety of robots come to market, with specific functions aimed at different needs. “Maybe 90% of robot applications can be covered by simple machines,” said Hirukawa, who points out that mobility, which can keep seniors independent for years longer, is the most important.

For example, Hirukawa, who’s worked on healthcare robots for over 30 years, thinks a “power suit” could solve the mobility issues that so many older people have. The suit is like an exoskeleton that can enable the wearer to move from sit to stand and to walk more easily. “The senior person can even sleep while wearing some of these power suits, so if they wake up in the night, they can get up and go to the toilet by themselves,” he said. This device is pretty simple, compared to a humanoid robot — it only has two motors and a sensor to sense the intention of the user — and it can learn and refine that feedback over time.

Atkeson agrees with the idea of taking the robotics out of the robot. “Let’s alter the home and roboticize the whole environment to help with the job [of elderly caretaking]. It’s not like we don’t do that now. In taking care of my own mother, we bought walkers and special pillows. Let’s put sensors in those to make them smart.” He envisions pillows that could help people off the couch and beds that work like Barcaloungers.

“We think we want a walking Swiss Army knife, but there’s no need to package everything we need in one robot,” said Atkeson. “Let’s take it apart and summon up the right tool to do whatever job needs to be done.”

Starre Vartan is an independent science journalist and dog-runner currently living on an island in Puget Sound near Seattle.