When Conor Walsh was a graduate student at MIT, he acted as a test pilot for his professor’s exoskeleton program. The powerful, rigid device was challenging to wear and work with, because of the way it had to interface with the body, forcing the user’s body to comply with the structure of the device, rather than vice versa.
Eventually, Walsh moved to Harvard, and started his own exosuit research. But he made a point to work on soft, pliable systems to aid locomotion. After five years of work, his suits have helped walkers move 20 to 25 percent more efficiently, according to his research, published recently in Science Robotics.
“The approach we’re taking, and a bunch of other groups are also starting to take, is can you deliver small to moderate assistance, but through a very lightweight and non-restrictive platform?” says Walsh.
The apparatus is based on a cable, which helps assist the motion of two different joints, the ankle and the hip. The user wears a harness around the waist, and straps extend from this harness to braces around each calf. A cable runs from the heel up to a pulley at the calf, and then off to a small motor. (For now, he has kept the motor and power source mounted elsewhere, as a way to simplify the study.)
Gyroscopic sensors mounted on the feet send data to a microcontroller, which interprets the walker’s stride and engages the motor at the appropriate time. As the motor reels in the cable, it tugs on the heel, aiding the step (called plantar flexion). The waist belt serves two purposes; it acts as support, so the calf doesn’t have to bear as much pressure, but it also offers assistance to the hip joint, as the force from the pulley is transferred upward via the straps.
Walsh and his coauthors ran the apparatus at four different power levels to see what was the most efficient.
“The main goal of this study was to look at, as we increase the amount of assistance we’re delivering to the person … what types of response do we see from the person?” says Walsh.
What they found was, even at the highest level of assist (measured by the force applied as percent of bodyweight, maxing out at 75 percent), they saw no plateau; efficiency, as measured by the amount of oxygen the participants used while walking, kept going up.
“What his data suggests is that, when you keep trying to add more assistance, there might be no limit, no boundary to how much we could improve a person’s gas mileage, if you will,” says Greg Sawicki. Sawicki also works in walking-assistive exosuits, as an associate professor of biomedical engineering at the University of North Carolina. His devices are based on a small, lightweight rigid exoskeleton—sometimes powered, sometimes actuated by a spring—that fit around the ankle.
“In our studies, we found a different result, which is that there’s often diminishing returns,” he says. “You do well up to a certain point of assistance, and then if you give too much, the efficiency of the human-machine system starts to decline.” He suspects some of the difference is due to Walsh’s multi-articular architecture, and how it incorporates the motion of the hip.
Both Walsh’s and Sawicki’s work has been applied to the medical field, helping stroke victims, or patients with multiple sclerosis, or other age-related injuries and disease to increase their mobility. Walsh has partnered with ReWalk Robotics to develop systems for these applications. But there’s a second important application, which has helped Walsh obtain DARPA funding: Soldiers schlepping heavy gear could one day use suits like these to help them walk farther, carry more, and experience less fatigue.
In pursuit of both goals, Walsh has been refining the textiles, the actuation systems, and the controllers to make such suits more realistic outside the lab. “The advances in this field are coming about through collaborations with people who understand the human, the physiology, the biomechanics, and people who understand robotics and the technology aspect,” he says. It’s a cross-disciplinary approach, featuring design and ergonomics, but also biomechanics, software engineering and robotics. Everyone walks a little bit differently, so the system must be at least partially customizable. And then there’s the weight.
“The biggest challenge is power density of the actuation,” says Sawicki, pointing out that mounting the batteries and motors on the walker instead of remotely on a nearby stand, as Walsh did, could decrease the efficiency. Until battery and motor technology improves, any increase in power requires an increase in weight, a tradeoff that is, for now, inherent in all such walkers. “There’s this fundamental rule that if you want to be more powerful, you gotta be heavier, when it comes to motors.”