Humanoid robots who can walk, run, and shake hands. A robotic hip exoskeleton to help stroke survivors rehabilitate their walking gait. A four-legged robotic guide ”dog” to help visually impaired people navigate the world independently. Robotic arms to perform tasks typically done by nurses.

These examples may sound like the stuff of science fiction novels, yet all are technologies currently in development at UMass Amherst. 

Image

Robotics research began at UMass Amherst around two decades ago, getting a major lift in the early 2020s when several new roboticists joined the faculty of UMass’s College of Engineering and Manning College of Information and Computer Sciences (CICS). These faculty members have collaborated across disciplines, including with colleagues in fields such as nursing and kinesiology, on research and innovation with applications for human health and health care. In 2023, they established the Robotics for Human Health and Mobility research theme within the UMass Institute for Applied Life Sciences’s Center for Personalized Health Monitoring. The group supports interdisciplinary research collaborations that aim to develop innovative robotics solutions to support human health, health care, and well-being.

Robots offer many advantages when it comes to health care and health research. Broadly speaking, robots can provide assistance (making tasks easier to improve quality of life), resistance (for training and rehabilitation purposes), and augmentation (extending human abilities), says Meghan Huber, assistant professor of mechanical and industrial engineering. Mobile robot systems, for example, can act as a second pair of hands for nurses and caregivers delivering patient care. Wearable robots, such as exoskeletons, can be used in assistance or resistance mode to improve walking for individuals with neuromotor or musculoskeletal impairments. At UMass Amherst, researchers are even exploring how robots can serve as advanced testing and intelligent sensing platforms for developing new medical devices. 

Yet, for all the hype that “machine learning will solve all our problems,” Huber adds, robotic systems are still in the early stages of development. Only within about the past decade have researchers created robots that can safely coexist with humans. 

“This opens up a lot of possibilities,” says Huber. “Now we’re at the hard part, which is getting robots to work well with humans and behave in ways that are actually useful.”  

Read on to learn how UMass researchers are working across disciplines to develop robotic systems to improve health care and enhance human health and well-being.

UMass Amherst Chancellor Javier Reyes shakes hands with a humanoid robot at Founders Day on campus, April 29, 2025.

Robotic Guide Dogs for the Visually Impaired

Image

Designing robots that perform functions useful for humans requires taking a human-centered approach to technological development. Donghyun Kim, assistant professor in UMass’s CICS, was inspired to develop a quadruped robot that could serve as a guide “dog” to lead blind or visually impaired individuals after he learned about a significant shortage of canine guide dogs. Working with a guide dog training center in Boston, Kim and his associates spent three years interviewing and observing visually impaired people in order to better understand their needs and how they interact with their guide dogs. The researchers published their findings in a paper that won a Best Paper Award at the 2024 ACM (Association of Computing Machinery ) Conference on Human Factors in Computing Systems (CHI), the leading venue for human-computer interaction research. 

“This initial research was a large push forward for the human-centered development of a robotic guide dog,” says Kim. “We are continuously getting user input as we develop new components for the robotic guide dog.”

For example, the researchers learned that because visually impaired individuals are highly dependent on being able to hear the world around them, it’s critical that a robotic guide dog can walk quietly. As a result, they found a solution that reduced the robot’s walking noise level by half.

Learn more about the robotic guide dog project in this video featuring PhD student Hochul Hwang.

Gait Rehabilitation for Stroke Survivors

Robots can also play a valuable role in physical rehabilitation for patients. Huber and Wouter Hoogkamer, assistant professor of kinesiology in the UMass School of Public Health and Health Sciences, are collaborating on several robotics projects that seek to rehabilitate walking function in stroke survivors. People who have had a stroke commonly experience neuromotor impairment in one half of their body, causing them to bear less weight on one foot and develop an asymmetric walking gait. Over time, this can cause other downstream problems at the musculoskeletal level, including osteoporosis and muscle atrophy, explains Huber.

From a kinesiology perspective, says Hoogkamer, therapists typically help patients improve their gait by physically manipulating their limbs or advising them to move their body differently during therapy sessions. But they know patients may stop moving as directed outside of therapy. “The robotic interventions we’re developing actually train patients to walk better in a ‘sneaky’ way, without telling them what to do,” he says. 

The robotic systems in development work by modulating the level of stiffness or compliance of the surface beneath one foot, which encourages the patient to walk more symmetrically. The UMass researchers have created a split-belt treadmill on which the stiffness of the walking surface can be adjusted independently on each side, as well as a shoe with an air pocket in the sole that allows the stiffness to be adjusted. 

Huber and her team have also created a robotic hip exoskeleton, which has shown promise to effectively train stroke survivors to modify their walking asymmetry. In January 2024, Huber, Hoogkamer, and their team published a proof-of-concept study in IEEE Transactions on Neural Systems and Rehabilitation Engineering, which demonstrated that applying resistive forces about one hip joint and assistive forces about the other with the exoskeleton mimicked the effects of split-belt treadmill training in neurologically intact individuals. The approach to the exoskeleton was inspired by the split-belt treadmill but allows for gait rehabilitation in real-world conditions.

“Because our exoskeleton is portable, it can be used during overground walking,” says Mark Price, a postdoctoral researcher in mechanical and industrial engineering and kinesiology and author on the paper. “We can build upon the successes of split-belt treadmill training with this device to enhance the accessibility of gait training and enhance the transfer of training benefits into everyday walking contexts.”

Another UMass team led by Frank Sup and Yahya Sadeghi-Modarres, professors of mechanical and industrial engineering, is developing a novel underwater gait training system, supported by the National Science Foundation. While hydrotherapy is commonly used for gait training due to its buoyancy and drag effects, this project introduces wearable hydrofoils that actively modulate lift and drag forces based on real-time control. Using a controller, the system determines the forces needed and adjusts the hydrofoil’s orientation to guide the wearer’s leg. This approach aims to provide compliant, energy-efficient assistance tailored to specific rehabilitation goals in aquatic environments, opening new opportunities for gait training systems.
 

Robotic Nursing Assistants

At UMass’s Elaine Marieb Center for Nursing and Engineering Innovation, which brings together the expertise of nurses and engineers, robotics are a major research thrust in the development of new health-care solutions. According to center codirector Sup, robotics are currently used in health-care settings to perform repeated tasks with precision, such as assisting with nursing duties, and have the potential to assist with many other tasks, including delivering supplies and medications, putting together medical kits, and disinfecting surfaces. Having robots take over this work would free up nurses’ time to interact more with patients.

Sup and his colleagues frequently partner with the nursing department at Baystate Medical Center in Springfield on robotic nursing technology research. Such collaborations are essential in ensuring that the systems developed are both effective in improving care and welcomed by patients and staff in health-care settings. 

Image

One current project focuses on improving robotic grasping of medical supplies, items that are often packaged in flexible, irregular, or transparent materials. The team is working with Cidália Vital, PhD, RN, at Baystate Medical Center, to develop a medical supply library: a dataset that captures images and data of actual hospital supplies to train robots for better identification and handling of these items. This foundational work will guide robotic platforms toward more reliable performance in real-world hospital environments.

To test these capabilities, the Marieb Center uses the Hello Robot Stretch, a compact, mobile manipulator about the size of an IV pole, to explore collaborative tasks with nurses, such as fetching items or assisting with simple physical tasks. Recently, the Marieb Center also acquired a Mobile ALOHA robot, an advanced four-arm platform capable of whole-body, bimanual manipulation and mobile operation. This system allows researchers to demonstrate complex, real-world tasks through teleoperation, enabling the robot to learn from nurse-led demonstrations. By studying how nurses perform tasks, the team is developing ways for robots to either support or autonomously complete elements of those workflows, with an emphasis on safety, usability, and clinical value.

“At UMass, we’re really focused on getting roboticists out of the robotics lab, collaborating with experts in different fields, and learning about real-world settings where robots can help solve problems,” says Sup.

Read More About the Elaine Marieb Center for Nursing and Engineering Innovation.
 

This story was originally published in May 2025.