A version of this story appeared in Science, Vol 375, Issue 6581.
In a sunny room at the Boys & Girls Club in Calistoga, a tiny city in California’s Napa Valley, Veronica Ahumada is setting up her robot when a bespectacled little boy wanders in to ask what she’s doing. Ahumada points to the device—basically a tricked-out tablet on a meter-tall post, mounted onto gyroscopic roller wheels—and he yelps like he’s seen a ghost. “She has a robot!” he hollers, bolting out the door.
Soon, about 40 tweens fill the room. Many are Latino like Ahumada, and they are here for after-school child care while their parents work. When she invites them to guess what the robot is for, their ideas bubble over: to play video games! To go to the supermarket! To help people get around! To take orders at a restaurant! To take over the world!
Ahumada, who studies health informatics and human-robot interactions at the University of California (UC), Davis, acknowledges each child’s suggestion, then explains: “I use these robots for children just like you,” but who can’t go to school because they have serious illnesses such as cancer or heart problems. Those children can send the robot instead, she says, and join their classmates virtually, from home. The group falls silent. “Whaaaaat?” they croon in a collective tune of disbelief.
In fact, rolling robot avatars allow ill children to attend school remotely in a few hundred classrooms across the United States and a smattering of other countries. The gadgets, called telepresence robots, were designed to allow doctors to conduct rounds from a distance and corporate executives to visit remote facilities. They show a remote user’s face and let the user see, listen, and speak as if they were present. Ahumada and a few other researchers think the devices could provide classroom access to a long-overlooked group of children. “We’ve never in the history of the world had the opportunity for children with these serious medical conditions to go to school with their peers,” Ahumada says.
Children who use the technology have called it life-changing because of the social connections it allows. “It’s like I’m actually there—that’s why I like it so much,” an 8-year-old told Ahumada. In addition to taking part in lessons as if they were in the classroom, children can zip around to chat with friends and join their peers for lunch, chorus, or recess.
“There’s probably a specific band of kids [for whom] this could be transformational,” says Gary Maslow, a pediatric psychiatrist at Duke University who studies how children adapt to chronic illness. But researchers haven’t yet proved these tools help academically, socially, or emotionally, he says. And the technology itself has limitations: The robots weren’t designed for children and don’t work well in schools with spotty Wi-Fi.
Ahumada is “trying to figure out how to overcome those obstacles,” says Justin Reich, an expert in educational technology at the Massachusetts Institute of Technology. Aiming to flesh out how best to integrate telepresence robots into classrooms, she and her colleagues have been probing how dozens of children with different illnesses, along with their families, classmates, and teachers, engage with the devices. In 2020, she teamed up with health care roboticist Laurel Riek at UC San Diego to design a machine with features tailored for children, including speakers that can carry sound over a classroom’s din, an “arm” for reaching and grasping, and a user interface operable by children with a range of ages and abilities.
She is also planning to join forces with clinicians to pin down whether the benefits that children report in her studies translate into gains in mental health, grades, or other measurable areas. If they do, she says, schools may eventually be required to provide robots, creating a market as well as a vision for the technology. “It’s much bigger than I originally imagined,” she says.
Over the past 2 years, as the pandemic turned the world inside out, children and families accustomed to attending school, sports, and scouts got a taste of isolation and homebound life. But by Ahumada’s count, about 2.5 million U.S. children with serious medical issues already experienced significant stretches of isolation before the pandemic. Teachers tell them not to worry about school and just to focus on getting better—a supportive sentiment, but one that underscores their exile from normalcy. Classmates often don’t know why they disappear from school or what their daily life is like.
Ahumada knows the experience well. Congenital heart defects that went undiagnosed until adulthood kept her home for weeks at a time during elementary school. She was too weak to do more than read or doze on the couch in her family’s one-bedroom home, waiting for her two brothers to come home with worksheets from her teacher. The days were lonely and achingly boring.
Children whose medical conditions keep them out of school are an amorphous group. Some get better and go back to school; others, sadly, succumb to their illness. Still others cycle in and out, as Ahumada did. U.S. public school districts generally provide such children several hours per week of instruction by a visiting teacher. But the system is spotty, and students often fall behind. According to research by Maslow and others, having a chronic illness in childhood puts people at a serious disadvantage. “They are less likely to attend college, half as likely to graduate college, less likely to be employed, and have lower income,” he says.
Children stuck at home also miss out on another aspect of school. “Humans have evolved to be social creatures,” says Maja Matarić, a roboticist and computer scientist at the University of Southern California. Social development goes hand in hand with cognitive learning, she says, and it benefits from a physical presence in school.
Ahumada was lucky: Her own illness subsided by middle school and her childhood normalized. She went to college and eventually landed a job at Montana’s Department of Public Health and Human Services. She soon found herself musing about technology and her solitary childhood. The department was expanding health care access by placing video consoles in clinics it was struggling to staff. Her colleagues expected patients would prefer in-person appointments, but later, when the state reverted to them, some people grumbled about losing the relationship with their distant doctor.
If patients can use video to bond with remote doctors, Ahumada thought, perhaps children at home could forge relationships with classmates and teachers the same way. “I wanted to know,” she says, “if it was good enough for physicians, was it good enough for kids? Do we have the technology to completely transform the daily experience of these children?”
Ahumada wanted to study the idea in graduate school, but she struggled to find an adviser. Eventually, she connected with Mark Warschauer at UC Irvine, who investigates how digital technologies can enable learning and social inclusion. His team wasn’t studying remote access for children, he told her—no one he knew of was—but he offered her a spot in his lab.
Colleagues were skeptical, though. One likened research on telepresence robots to “studying TV carts.” Another told her the work would have no impact because “so few” students have illnesses that require them to stay home. Even Warschauer didn’t seem fully convinced, she says, until he saw a Verizon ad from the 2013 Super Bowl, which featured a child maneuvering a robot through school from a hospital bed. “Mark had never seen my idea conceptualized before,” Ahumada says. “We were so excited—we were like, ‘It’s a thing!’”
The device in the ad was manufactured by VGo, a small company that had already begun to sell robots to schools for ill students but hadn’t done research to discover how well the devices met their aims. (The company has since been acquired by Vecna Technologies.) Ahumada contacted VGo, which connected her with a school district in Texas that had just purchased several robots. The program had started with a girl in elementary school who was using a robot while undergoing cancer treatment. Her classmates were so grateful their friend could remain among them during her illness that they raised more than $1000 to help buy another robot for other ill children to use.
Ahumada decided to base her first case study there, using interviews to probe how ill children and their community used the devices. The logistics were complicated. Because the robot captures the classroom on video, families of every student had to consent to its presence. Recruiting robot users was another challenge. Like most families with a seriously ill child, the ones in this district “were just trying to survive,” Warschauer says. “They weren’t thinking about participating in research projects.” But by the end of 2013, Ahumada had completed the study, interviewing five sick students from second to ninth grade about their experiences with the VGo robots, as well as five parents, 10 teachers, 35 classmates, and six school administrators.
The ill students all told Ahumada about relief the robot brought from social isolation. One child’s mother said she didn’t realize her son was depressed until she saw him blossom with the robot, spending much more time alert and engaged with school. The study also revealed classmates and teachers quickly came to treat the robot not as a moving hunk of plastic and metal, but as the student it represented. Children and adults referred to it by the student’s name, and many classmates went out of their way to help when the robot got stuck. “My research kind of gives me hope for humanity,” Ahumada says. “The majority of kids are so thoughtful.”
Not every experience was positive. A ninth grade girl decided to return the robot because of the unwanted attention it brought—classmates teasingly called her a “vacuum cleaner,” for example. And a fifth grade boy was bullied by a classmate who kept smearing the robot’s lens with ketchup.
Matarić, too, has explored how the technology works in schools, using robots made by a company called Ohmnilabs. She and her colleagues first had design experts operate a classroom robot remotely, as a child would, to identify technical features that would help children and teachers use them more effectively. These included a signal like raising one’s hand to get a teacher’s attention and a camera that swivels to look at a classmate. A study that gave robots to four children to use in their classrooms over 2 to 8 weeks also identified the need for a speaker that can regulate the volume of its voice according to whether a child is whispering questions about English homework or being chased on the playground.
All four children found the experience very positive, the researchers saw, but deploying the robots involved extensive coordinating and troubleshooting, as well as navigating school district politics. They weren’t ready to use out of the box, Matarić says.
Ahumada’s work similarly finds a mix of promise and complexity, as have the few other studies that have looked at telepresence robots used in schools. Since her pilot study in Texas, she has gathered perspectives from 91 children with illnesses across the United States who have used telepresence robots, along with their families, classmates, teachers, and school administrators. The important design needs those interviews revealed identify some of the same features that Matarić pinpointed. But the children’s most common demand was for a robotic arm that would allow them to engage with the world—reaching, grabbing, tagging. Children want something that enables “not just seeing and hearing and moving around on your own, but also being able to touch the world and to receive sensations,” Riek says.
The children also voiced values that might appear frivolous but could help them maintain social connections. One child requested a robot feature that’s sure to be a design challenge—the ability to do a bunny hop. And 53 of a subset of 82 children Ahumada interviewed described using the robot to play, either with classmates or alone, she and Riek report in a paper under review. One child told Ahumada that using the robot showed their friends they remained the same person they were before they got sick. Peers, too, seek that connection. “Little kids hug the robot because they’re so happy to see their friend,” Ahumada says.
As the world enters the third year of the pandemic, remote attendance through video conferencing platforms has become routine. That change may actually harm the prospects for telepresence robotics in schools, Ahumada says. Video conferencing is cheaper than robots, and teachers now have experience using it. When several schools she works with reopened, they chose to leave students who were remote for medical reasons on video rather than bring back the robots.
The pandemic has also revealed an important limitation of virtual learning, Reich notes. “Teachers have a very hard time managing classes when some people are in person and some people are at a distance,” he says. The difficulty may persist even with a single remote student on a robot.
But better telepresence technology could help, Reich adds. “I could imagine some future where these things are way better than Zoom.” What if, he wonders, “holding a virtual reality controller, you point your finger at something, and the robot takes its finger and points at it?”
In September 2020, Ahumada and Riek received a $1.2 million grant from the National Science Foundation to develop a telepresence robot with features specifically designed for remote learning for children. They are starting with a prototype called Stretch, made by Hello Robot, that was designed for older adults and people with disabilities to use around their home. For now, Stretch lacks a display screen, but it has a retractable arm that moves smoothly up and down a sturdy metal spine. At the end of the arm sits a gripper consisting of two small rubber cups on bendy strips of metal.
In Ahumada’s office, a graduate research assistant, Jingjing Xie, explores Stretch’s capabilities. Sitting on a chair with the classic board game Trouble at her feet, she uses the robot’s controller to bring the arm down to the board and move a yellow peg. It’s clumsy, like a claw machine at an arcade, and the gripper’s rubber cups can’t press the popper that rolls the dice.
Stretch is a work in progress. Over the next year, based on Ahumada’s studies to date, she and Riek will settle on features that a child needs to feel present and engaged in learning, and they will modify Stretch to include them. Remote students will test their prototypes and weigh in. “These kids are the pros,” Ahumada says. Long before the pandemic, before many families gave virtual school a thought, “they were already the pioneers.”
Some alterations should be straightforward, such as adding a screen to show the face of the child at home. Other specs will surely require inventing technology. And the researchers want to make the robot’s physical presence cool and fun for children, something they can feel proud of as an extension of themselves among their peers. “We want the child to feel like they’re not this weird thing inside a robot, but they are part of the class,” Riek says.
Still, despite Ahumada’s investment in telepresence robots, she would happily abandon them if something more nimble came along. “Today it’s robots,” she likes to say, “but tomorrow it could be holograms.” It’s all in support of a bigger goal. When she visits classrooms with a robot in play, she sees sick children mentioning their nausea or showing classmates a port for delivering medicine or a scar from a recent surgery. Classmates breezily ask their robot-embodied friends how they are doing. Healthy and sick children form a single community. “If we don’t scoot illness away,” she says, “will we be more compassionate?”