In Conor Walsh’s engineering lab at Harvard University, no one looks askance at a staff member wearing a loudly whirring backpack, with wires snaking out and down his leg. A trio of sewing machines have their own workroom. A dozen pairs of identical hiking boots neatly fill a shoe rack on the far side of a treadmill. A disembodied glove clenches and straightens as air fills and drains from its fingers.
All of this equipment is aimed at helping people move faster, more smoothly, while expending less energy. Walsh, also a core faculty member at Harvard’s Wyss Institute for Biologically Inspired Engineering, is most excited about the devices his group is designing for stroke patients, who often struggle to regain their strength and fluidity of movement. The team has already shown that “soft exosuits” can provide a robotic assist to movement, enabling soldiers to march for longer.1 This month, the researchers will begin clinical trials to test the suits’ ability to help stroke patients relearn how to walk efficiently.
Walsh belongs to a growing group of researchers worldwide who are using small, lightweight robotics to help people with a range of medical conditions that hinder mobility. Rigid, whole-body “exoskeletons” have made headlines in recent years—perhaps most famously when a suit developed by Duke University neuroscientist Miguel Nicolelis and colleagues enabled a 29-year-old paraplegic Brazilian man to kick a soccer ball at the launch of the 2014 World Cup in São Paulo. Such exoskeletons have helped paralyzed people to walk again, albeit awkwardly, by pushing, pulling, and supporting them to stand up and move one leg followed by the other. But Walsh and other researchers have realized that people with less-disabling conditions need a subtler boost. Now, teams around the world are developing smaller, lighter devices that help, rather than drive, movement. And sewn into clothes, they can be donned as easily as pulling on pants, a shirt, or gloves.
A group in Italy is designing a suit to reduce falls among the elderly and amputees, for example, while researchers at Stanford University are trying to reduce the energy it takes people such as those recovering from stroke to walk. And a team in New York City is helping children with cerebral palsy get out of the “crouch gait” that makes it difficult and awkward for them to get around.
There’s a general appreciation both within the scientific and clinical community that these robotic devices can make a big difference in people’s lives.—Sunil Agrawal, Columbia University
These advances are supported by a number of technological improvements and cost reductions over the past decade, Walsh says. Motors are smaller, more powerful, and cheaper. Electronics are easier to use. Gyroscopes and accelerometers are now so tiny, inexpensive, and precise that they can give directions on cell phones—and can tell precisely where someone’s leg is in space and what direction it’s moving in. “The technologies that robotics research groups can pull from have gotten better across the board,” Walsh says.
“It’s a great time in the field,” agrees Columbia University’s Sunil Agrawal, who leads the work on cerebral palsy. “There’s a general appreciation both within the scientific and clinical community that these robotic devices can make a big difference in people’s lives.”
Walsh’s work on exosuits started nearly six years ago as a collaboration with scientists at the US government’s Defense Advanced Research Projects Agency (DARPA), aimed at reducing the energy soldiers have to exert to carry heavy backpacks over long distances. (See "Beyond the Clinic") But a few years into the project, the researchers began to realize the technology’s potential for helping patients, too—in particular, people recovering from stroke, which affects nearly 800,000 Americans a year, leaving many with physical disabilities.
If they’re both lucky and well-insured, stroke patients get a few weeks of inpatient rehabilitation therapy, says physical therapist Terry Ellis, who collaborates with Walsh and directs the Center for Neurorehabilitation at the Boston University College of Health and Rehabilitation Sciences: Sargent College. But with limited time, rehabilitation specialists focus on getting patients walking again in whatever way possible, often with the use of a walker, a cane, or a hard plastic 90-degree brace that keeps their weaker foot from “dropping” as they lift it off the ground to take a step. Many patients never learn to walk normally again, Ellis says. And because the plastic brace keeps the patient from being able to push off the ground with that foot—an essential part of the biomechanics of walking—the more that person walks, the weaker the ankle gets, and the more the foot drops, she adds. “We’re missing out. We’re not optimizing on the potential people have to improve.”
Stroke patients in wheelchairs fall even further behind: not only do they lack support for working on walking skills, but constant sitting impairs bowel and bladder function, reduces bone mass, and dysregulates blood pressure, notes Paolo Bonato, a researcher at Spaulding Rehabilitation Hospital in Boston who is collaborating with Walsh and Ellis. “Being in a standing, load-bearing position is actually quite important for the body,” he says. And a soft exosuit may be just what stroke patients need to get back on their feet.
In Walsh’s lab, graduate student Jaehyun Bae dons a version of the device the group has developed and takes to the treadmill. As he pretends to walk with a dropped foot, a wire from the device wrapped around his calf and ankle pulls up his foot at just the right second to avoid hitting the floor, then quickly lets it go so he can push off. When he picks up his pace, the robotic movements speed up with him. Bae then shifts his gait to swing one leg outward. Again, the device matches his stride to pull the leg back in line.
In preliminary tests conducted by Walsh’s collaborators at two Boston clinics, the device seems to be helping stroke patients. Not only does it appropriately correct for the users’ aberrant movements, it helps increase their pace. A healthy young adult generally walks about 1.2 meters per second, Walsh says. Someone who walks slower than 0.4 meters per second is considered essentially homebound; those with a pace of 0.4–0.8 meters per second can get out occasionally, and those whose speed exceeds 0.8 can fully integrate into society. “If you could get someone from 0.3 to a 0.6, or a 0.6 to a 0.9, that would be a big deal,” Walsh says. Last July, he and his team showed substantial progress toward that goal.2 “We’re not talking about tremendous changes, we’re just trying to give enough of a little boost to push people over these thresholds so they can start to be more active.”
Early this year, in collaboration with ReWalk Robotics, Walsh and his colleagues will evaluate a commercially viable version of the exosuit for stroke patients. They hope to win US Food and Drug Administration (FDA) approval by the end of 2018. Concurrently, the researchers are developing a system that works on both legs that should be ready within a couple of years to help people with a greater range of ailments, including multiple sclerosis, cerebral palsy, ALS, and Parkinson’s disease, Walsh says. Further off but also under development: devices that will address arm issues in the same diseases, he adds. “We’ve been starting to test and starting to understand how we can best help someone who has an upper extremity impairment.”
Robotic help for a range of conditions
A handful of other research teams are also developing soft exosuits for patients with movement disorders. In Pisa, Italy, for example, Vito Monaco at the Scuola Superiore Sant’Anna has developed a pelvic support system for the elderly and amputees that, at least in the lab, can help right someone before they fall.3 “It’s not easy to predict the way a person will fall down,” says Monaco. “We as roboticians should combine detecting falls or lack of balance with strategies to assist people to recover their balance.” A Belgian group at Ghent University is also aiming to help the elderly get around, and just last year the researchers showed that the exosuit they built to assist with plantarflexion—the “push-off” stage of walking—can reduce the effort it takes for a person over the age of 65 to walk.4
Meanwhile, Columbia’s Agrawal focuses on in-hospital patient rehabilitation and training, using cable-driven exosuits to train children with cerebral palsy5 and adult stroke patients to improve the coordination of their limbs. Instead of a single wire or set of wires to help someone who drags a foot, Agrawal says his group’s exosuits, which are often attached to the ceiling for stability, have multiple wires that apply forces to manipulate the gait in a much more fine-tuned way appropriate for rehab therapy. He is particularly interested in posture and the curvature of the spine, and has written several recent papers showing that his team’s exosuits can train patients to correct for postural weakness.6
Other projects around the world include an effort at the University of Michigan to develop a robotic ankle that adapts to the gait of its user, currently being tested on healthy people,7 and collaborative work at North Carolina State University and the University of North Carolina at Chapel Hill meant to help the mechanics and control of ankle muscles and tendons.8 Meanwhile, Walsh’s graduate advisor, Hugh Herr at MIT, is working with colleagues to develop bionic prosthetics and exosuits meant to help amputees as well as healthy people and other patients hop, run, and walk.9
Even big-name corporations have expressed interest in the field. Samsung, for example, is developing full-body and hip-only exosuits designed to support walking in the elderly and disabled, and eventually to improve performance in soldiers. Honda has been developing an assistive exosuit for people with total paralysis. And Toyota announced earlier this year that its rehabilitative exosuit, aimed at people with lower-limb paralysis, would soon be available for rent by medical facilities.
The road ahead
These technologies still face their fair share of challenges. In fact, there are only eight groups around the world “that have demonstrated a device that can improve performance of the user,” notes Steven Collins, an associate professor of mechanical engineering at Stanford University who recently moved there from Carnegie Mellon. “And all of those have been demonstrated in the last four years.”
We’re at the Toshiba level right now. We have a really long way to go before we get to a MacBook Air type of system.—Paolo Bonato, Spaulding Rehabilitation Hospital
One problem is practicality. Monaco, for example, is still struggling to make his exosuits light enough so that they don’t further destabilize users. “The last version of our [robotic] pelvis weighs 3 to 4 kilos, which is quite a huge backpack for an elderly person—more than a couple of big bottles of water,” he says. Bonato agrees, comparing the exosuits of today to the laptops of 10–15 years ago. Back then, his Toshiba computer was ostensibly portable—but just barely. “We’re at the Toshiba level right now. We have a really long way to go before we get to a MacBook Air type of system.”
And it’s not just the size of the devices, but their function as well, Collins adds, arguing that most of the failures stem from a lack of understanding about how best to help. “It’s really easy to accidently make it harder for a person” to walk, he says.
His and Walsh’s groups are now employing an iterative approach, in which devices can be changed or “learn” as people interact with them. Too many research groups focus on making a generic device that will move everyone’s legs, without addressing the individual motions that make up that movement, Collins argues. With an iterative approach, “you can try lots of designs really quickly without having to build new, specialized hardware” for every person who uses the device, he says. Using an optimization algorithm to efficiently explore the potential movements that might help, Collins and his colleagues showed they could reduce energy expenditure by 24 percent with an exosuit tuned to a healthy individual—an improvement four times greater than they’d ever achieved by making the variations by hand.10 “We were floored,” he says.
Collins says that making a device that will actually help people requires involving them in the development process. The hardware and software are important, and so too is the person inside the suit, he notes. “When we started optimization work, we thought the most important thing was to find the [best] device. [But] just as important as the device learning the person is the person learning the device.”
Karen Weintraub is a freelance science writer living in Cambridge, Massachusetts.
A VIABLE MARKET?
He says he’s hoping that within a few years he will be able to offer even less-expensive suits tailored for other patients: people suffering from multiple sclerosis or Parkinson’s disease who need even lighter movement nudges. Jasinski says he expects to round out the offerings with exosuits aimed at people with cerebral palsy and the elderly, once the suit can be redesigned and tested for them. “If we can handle all those, we’ve got a very large industry,” he says.
These newer devices would cost a lot less—on the order of just $19,500—than the whole-body exoskeletons for paraplegics, which run about $80,000, says Jasinski, who notes that rehabilitation hospitals could be financially motivated to buy the suits because fewer staff members will be needed to work with each patient. At-home costs may be harder to justify to an insurance company, at least at first, he says. But a patient might be able to rent the suit as needed, to bring the price down to a manageable sum. “I can make that work from a business model.”
BEYOND THE CLINIC
The robotic exosuit is intended to help soldiers march faster and farther while carrying heavy packs, Eckert-Erdheim explains. The latest version of the Harvard suit weighs 4.5 kg and can reduce a soldier’s effort by 5 percent to 10 percent in real-world situations. Eventually the goal is to achieve a 25 percent energy reduction, says Walsh, a faculty member at Harvard’s Wyss Institute for Biologically Inspired Engineering who in 2014 received a $2.9 million contract from the Defense Advanced Research Projects Agency for the work.
Back at Harvard, Walsh and his team are still meeting with Army officials to figure out how they might proceed with the project even though the lab’s last military contract has ended. “We don’t have a clear idea yet, but we’re excited about talking to the Army and the military medical community,” says Ignacio Galiana, a staff robotics engineer at the Wyss Institute and former postdoc in Walsh’s lab.
First responders such as firefighters could also benefit from devices that reduce their fatigue as they climb flights of stairs or carry limp bodies, for example. Researchers also see a market for these exosuits in athletics, particularly when it comes to training, “to sense how you’re doing and give you feedback on how you could improve,” says Galiana. “Or you could imagine wearing the device to improve your performance or to recover faster after you do a lot of exercise” or suffer an injury.
Correction (Feb. 2): The original version of this story incorrectly identified MIT's Hugh Herr as Conor Walsh's postdoctoral advisor. He was, in fact, Walsh's graduate advisor, as indicated in this corrected version. The Scientist regrets the error.
- Y. Ding et al., “Biomechanical and physiological evaluation of multi-joint assistance with soft exosuits,” IEEE Trans Neural Syst Rehabil Eng, 25:119-30, 2017.
- L.N. Awad et al., “A soft robotic exosuit improves walking in patients after stroke,” Sci Transl Med, 9:eaai9084, 2017.
- V. Monaco et al., “An ecologically-controlled exoskeleton can improve balance recovery after slippage,” Sci Rep, 7:46721, 2017.
- S. Galle et al., “Exoskeleton plantarflexion assistance for elderly,” Gait Posture, 52:183-88, 2017.
- J. Kang, et al., “Robot-driven downward pelvic pull to improve crouch gait in children with cerebral palsy,” Science Robotics, 2:eaan2634, 2017.
- M.I. Khan et al., “Enhancing seated stability using trunk support trainer (TruST),” IEEE Robot Autom Lett, 2:1609-16, 2017.
- J.R. Koller et al., “Learning to walk with an adaptive gain proportional myoelectric controller for a robotic ankle,” J Neuroeng Rehabil, 12:97, 2015.
- K.Z. Takahashi et al., “Adding stiffness to the foot modulates soleus force-velocity behaviour during human walking,” Sci Rep, 6:29870, 2016.
- H. Herr et al., “Bionic ankle–foot prosthesis normalizes walking gait for persons with leg amputation,” Proc Biol Sci, 279:457-64, 2012.
- J. Zhang et al., “Human-in-the-loop optimization of exoskeleton assistance during walking,” Science, 356:1280-84, 2017.