That humanoids and humans will someday coexist in close proximity is looking less like science fiction and more like a near-future certainty. But just how adaptable are we to each other? That is a question 39 researchers from six universities and multiple industry partners will take up through a new $30 million, five-year National Science Foundation-funded Center for Human and Robot Co-Adaptation, led by the University of Texas at Austin.

The undertaking is aimed at something far more complicated than teaching robots to perform household chores. Researchers want to understand what happens after robots move out of carefully controlled laboratories and into homes, hospitals, dormitories and assisted-living facilities, where human behavior can be unpredictable and needs can change from one day to the next. The center will study what researchers call “human-robot co-adaptation,” the ways people and robots learn about each other and gradually adjust their behavior while sharing the same spaces.

Think about a comfortable living room, with chairs and tables arranged around a rug, decorations on side tables and perhaps a glass of water sitting just beyond someone’s reach. Now imagine a large robotic arm on wheels rolling into the room, extending toward the glass, grasping it and delivering it to a person sitting on the couch. That is fairly basic stuff at this juncture in robotics. What researchers want to understand are the subtleties that come next: the individual, the space, habits, preferences and physical limitations, and whether a robot can learn all of them.

Perhaps the person on the couch has limited mobility and can only comfortably reach so far. A useful robot would eventually need to recognize that and adjust where it holds the glass. More importantly, it would need to remember. As the person becomes familiar with the robot, he or she might also discover new ways to use it. The machine learns the human, while the human learns the machine.

“The next leap in robotics is not just getting robots to perform more tasks,” said Joydeep Biswas, a University of Texas associate professor of computer science who will direct the center. “It is enabling robots to understand the people around them: their needs, preferences, values and constraints, while recognizing that people also adapt their behavior when they integrate robots into their environment.”

Most robots are designed and tested around a fairly straightforward goal: successfully complete a particular task. But success in a home is more complicated than success on an assembly line.

The center’s researchers will explore whether robots can learn new skills, discard outdated ones, anticipate changing needs and recognize some of the unwritten social rules humans absorb almost without thinking. In a sense, the work resembles deep learning on an intensely personal scale, with machines learning individuals over time and people simultaneously learning the capabilities, quirks and limitations of the machine.

The center’s Human Environment with Robots, or HERO, Facility Network will include houses, dormitories, cafés, a public museum, a rehabilitation hospital and elder-care and assisted-living residences. At the University of Utah, researchers plan to deploy mobile robots at University of Utah Health sites, including the Craig H. Neilsen Rehabilitation Hospital, initially performing relatively simple fetch-and-delivery tasks before potentially moving into cleaning, tidying, assistive feeding, social interaction and navigation.

“A lot of research into robotics today focuses on whether a robot can successfully complete a particular task,” said Daniel Brown, an assistant professor in the University of Utah’s Kahlert School of Computing. “Through this center, we want to dive deeper and understand what happens when a person and a robot live and work together for weeks, months or even years.”

That long view could be consequential for people with disabilities or those undergoing rehabilitation. The Neilsen Rehabilitation Hospital already has 75 smart patient rooms where assistive technologies are incorporated into care. Researchers will be able to study not only how robots respond to patients but how patients, caregivers and medical professionals change their own behavior as the machines become part of their routines.

“This project gives us a chance to study robotics where the technology really matters: alongside people in their everyday environments,” University of Utah researcher Jason Wiese said. “We want to understand how robots can adapt to individual needs and support greater independence.”

The center combines expertise in AI, robotics, cognitive science and social science. Along with UT and Utah, participating universities include the Massachusetts Institute of Technology, Yale University, Indiana University Bloomington and Tufts University. Industry collaborators include Amazon, Apptronik, Diligent Robotics, Google DeepMind, Hello Robot, MassRobotics, NVIDIA and Robust AI. UT’s Texas Advanced Computing Center will provide computing resources for simulations, machine learning, data storage and digital replicas of real-world sites, allowing researchers to test scenarios virtually before trying them around people.

Introducing machines capable of learning human habits into private and semi-private spaces raises questions about consent, privacy, trust and control. Community members will have opportunities to influence where and how robots are deployed, according to UT, while an internal ethics board will oversee the research and develop procedures allowing people to consent or opt out.