Robotic-powered prostheses – state of play

T

he idea of a robotic-powered limb may sound like something from a sci-fi film. The phrase conjures up images of superintelligent cyborgs, in which humans transcend the limitations of biology by merging with machines.

While that scenario is some way off for now, the bionic limbs themselves are more than a pipe dream. Increasingly, prosthetics are being integrated with advanced robotics, with a view to restoring the functionality of a healthy human limb.

These devices are enabling patients to regain their motor capabilities and even restore sensory feedback – and many of the devices under development are mind-controlled.

“Only a few years ago, robotic prostheses could only do basic tasks, like standing up and walking at a constant speed indoors,” says Professor Tommaso Lenzi, director of the Bionic Engineering Laboratory at the University of Utah. “Now they can perform many complex movements, similar to biological legs. They can walk on inclined or rough terrains, climb stairs, squat, lunge, and even run.”

The need for these kinds of devices is not in doubt. Millions of people around the world live with limb loss, often due to diabetic neuropathy or trauma. Around 185,000 amputations are performed annually in the US, along with 431,000 in Europe.

For these patients, their quality of life will depend to a large degree on the quality of their prosthetic devices. Traditional passive prostheses may give them some mobility, but they can’t restore natural muscle function and often end up restricting the person’s activity levels.

To date, most robotic-powered prostheses are in the early stages of development, and the research prototypes have yet to be translated into commercial applications. However, once they do reach clinical viability, they could prove genuinely life changing.

This is a complex and multidisciplinary field, combining neuroscience, robotics, computing and medicine among others. At the AMBER Lab at Caltech, scientists are using mathematical algorithms to control robotic devices.

“We leverage the underlying mathematics of locomotion together with representations of musculoskeletal models of humans,” says Professor Aaron Ames, who runs the AMBER Lab. “This allows for the automatic generation of dynamic walking gaits that can be directly realised as algorithms on robotic assistive devices.”

Meanwhile at the Neurobionics Lab at the University of Michigan, researchers have a dual mission. As well as developing wearable robotic technologies, they are working to advance our understanding of biomechanical science – asking questions about how limb mechanics are felt and regulated by the nervous system.

“One thing we do is to study the properties of the human body during gait,” says Dr Elliott Rouse, professor of mechanical engineering and director of the Neurobionics Lab. “This involves perturbation of the joint, measuring its response and then characterising the stiffness and damping properties. We use that to guide the development of assistive technology.”

On the tech side, the lab has several projects underway, including an open source robotic leg. This leg can be easily manufactured, assembled and controlled, so that research teams wanting to study these systems don’t have to build a device from scratch.

They are also working on a variable stiffness foot and ankle prosthesis, which can change its springiness and shock-absorbing properties from step to step. In other words, it adjusts its behaviour depending on the conditions or gradient, just as a real ankle would.

“That’s potentially the holy grail of intelligent prostheses, because it allows you to not only vary the behavior of the foot for different activities, but also vary the energy storage so the person gets the kinetic energy returned to them,” says Rouse.

Classically, prosthetics were controlled by moving other parts of the body, or through toggling switches and buttons. With more advanced prosthetics, the device is integrated with the nervous system, and controlled via electric signals from the nerves and muscles above the amputation.

In 2020, an international team of researchers reported on a mind-controlled arm prosthesis that delivers the sensation of touch. The device is directly connected to the user’s nerves, muscles and skeleton, and generates sensations that are perceived as arising from the missing hand.

More recently, a team led by the Johns Hopkins Applied Physics Laboratory reached a milestone in their research – a partially paralysed user was able to feed himself using a pair of robotic arms. The prosthetics were integrated with a brain-computer interface, allowing the man to maneouvre the arms naturally and without much mental input.

The study built on the Revolutionizing Prosthetics programme, which was originally funded by the US government. This programme aims to create neurally controlled arm prosthetics with near-natural motor and sensory capabilities.

Generally speaking, upper limb devices are evolving rapidly, and a number of bionic hand and arm prostheses have been rolled out commercially. Esper Hand, for instance, is a self-learning bionic hand prosthesis that gets faster and more precise the more it’s used.

Despite the buzz around bionic arms, Rouse cautions that lower limb prostheses are lagging behind. Although there are some robotic leg prostheses already on the market – one ankle and one knee – they have not seen the clinical impact one might have hoped. Compared to arm prostheses, these devices are much harder to control.

“The field has remained mostly the same for the last five to 10 years, because telling robotic prosthetic legs what to do is extremely challenging,” says Rouse. “If they make a mistake on a set of stairs, the risk is really high, because the legs support the body. The instructions provided to these systems need to perfectly match what the user is trying to do at all times, and that's a really high bar.”

He adds that, as things stand, robotic prosthetic legs use no information from the human body – the cost of a misunderstanding is too high, and so all the information comes from sensors on the device. Over the long-term, that needs to change.

“What I think our field really needs is a direct neural connection to control these prostheses, either from the central or peripheral nervous system,” says Rouse. “These will be cortical implants or implants on the peripheral nerves that tell us more clearly what the human body is trying to do. There are researchers around the world who are looking at that.”

For instance, Lenzi’s lab in Utah is seeking to develop more intuitive control strategies for prosthetic legs.

“We are using different approaches to read the neural signals generated by the user’s nervous system and translate that into voluntary movements of the prosthesis,” says Lenzi.

Apart from the control issue, these systems face other problems too. One big issue is weight – most robotic prostheses are still too heavy to be used every day. Another is power. Not too long ago, the devices needed to be tethered to a power supply, preventing their application in real-world settings. Even today, they have limited battery life, typically a few hours at most.

“This is not acceptable and most people would want a prosthesis that can work a full day before needing to be recharged,” says Lenzi. “Unfortunately, using bigger batteries is not an option because they will add weight, which is not tolerable.”

Yet another issue is noise – robotic prostheses are much louder than their conventional equivalents – and still another is intuitiveness. As Lenzi points out, they are still quite hard to use, and generally only younger, stronger amputees are able to use them to their full potential. For his research team, these kinds of challenges represent an open goal.

“At the Bionic Engineering Lab, we use an interdisciplinary, holistic research approach to develop robotic leg prostheses focused on the user's needs,” he says. “Over the last few years, we have developed new actuation technologies that break the traditional trade-off between mass and power of conventional systems, allowing for robotic leg prostheses that are lightweight, strong, and efficient.”

For Rouse, the goal is simply to bring his existing projects to fruition – guiding early-stage projects towards clinical viability, and clinically viable projects towards market approval.

“We've come a long way, and we've done a lot of really interesting work that's advanced our field, but still patients don't see it,” he says. “Our work in the future is doing the due diligence in science and engineering, to get this work to patients who need it.”

Lenzi adds: “Robotic leg prostheses have the potential to help millions, but we need to move them from the lab to the real world, to home and the community where people need them the most.”