Over the past few years, we've heard some vague suggestions that powered exoskeletons will, “at some point soon,” be available for applications besides meeting the needs of people doing physical rehabilitation, industrial workers, and the elderly. The rest of us could get plenty of use out of exoskeletons too, for any situation in which our bodies need to support more weight for longer than would otherwise be comfortable. It's understandable that most exoskeleton companies aren't going for the consumer market right away, because exoskeletons are expensive, and folks like you and me simply wouldn't be able to justify the cost.
Yesterday, Roam Robotics (which was birthed from company-known-for-doing-weird-things, Otherlab) announced that it's planning on selling a new kind of exoskeleton designed to offer leg support to skiers and snowboarders. The cost? Just $2500. The gadget, powered by soft pneumatic actuators, will be rentable at ski resorts for way, way less than it’s full purchase price. Roam says that without any training at all, it'll enable you to ski better for longer without getting nearly as tired.
Pneumatic actuators have lots of advantages, both for exoskeletons specifically and robotics in general. Relative to electric motors, they're super cheap, and super lightweight. The big problem with them has always been the need for a portable energy source (batteries) and then a separate system to generate, store, and transport pressurized air. Electric motors, on the other hand, need only batteries and motors. And for most robots (especially the ones where price is no object), that setup is more efficient. Electric actuators can also go forwards and backwards, while pneumatic actuators work in opposition like muscles do, meaning that you effectively need twice as many of them to achieve the same degree of freedom. Because of this, we've mostly seen pneumatic actuators being used in applications where softness is particularly important, like with robots that are squishable, because there's no way to make an electric actuator soft.
Rather than focusing on the softness aspect of pneumatic actuators, Roam is instead leveraging the order of magnitude cost and weight reduction that the give you relative to electric motors. Overall, the pneumatic system is a bit more complex, and it's not going to give a skier the same performance as an electric motor would. But in an assistive exoskeleton application, you may not care about those things nearly as much as you do about weight and cost.
Photo: Roam Robotics
The way Roam Robotics' exoskeleton works is simple. It's essentially a shock absorber that takes over some of the work you'd otherwise be doing with your quads. As you bend your knees, pressurized air is pumped into the nylon actuator, which expands to provide some counterforce between the ski boot that the exoskeleton is attached to on one end and your quad on the other. Long story short: Your muscles have to do less work. For older folks, the system will help them ski better even if their strength and stamina are limited. Younger skiers will be able to go down the slopes faster and more aggressively. Sensors allow the exoskeleton to do its thing autonomously, gradually learning how you ski and providing the right amount of assistance at just the right time. In total, the current system weighs 9 kilograms (20 pounds), including both the device itself and the backpack that holds the batteries and pneumatic system—although that could decrease by as much as 50 percent.
Skiing might seem a little bit random as a first application for an assistive exoskeleton, but I'm guessing there are a bunch of good reasons they went in this direction. First, you want to start somewhere that people won't have to buy one without trying it first, which means renting it out from a centralized location, like a ski resort. Second, the task has to convey a very near-term positive benefit— not just "you're less likely to get hurt," but "you will perform noticeably better and feel better doing it," which is an experience that folks seem to have while using the exoskeleton. Third, you want to minimize the inconvenience of having to carry batteries and all the rest of the pneumatics, but many skiers wear backpacks anyway, so that's not a big deal. Fourth, for an exoskeleton to be weight bearing, there needs to be a way of transferring that weight to the ground that avoids your legs. Other exoskeletons use special rigid shoes, but skiing is perfect, because everyone is wearing rigid boots already. And lastly, I would guess that the demands on the system for skiing and snowboarding tend to be relatively predictable, since motions (from people who know what they're doing) are rhythmic, with smooth increases and decreases in how much support the actuators are asked to provide. This puts less strain on the system, and likely makes the control problem somewhat easier.
This is not to say that Roam Robotics is stuck with just skiing and snowboarding as applications. The company is likely expecting to learn an awful lot from this first real-world use case, and once the technology is proven, it'll be able to expand relatively quickly into applications like lift assistance for industry and commerce, or support for the elderly to help them with sit-to-stand transitions as well as stair climbing. And maybe if you need to move some furniture, you'll be able to go down to your local hardware store and rent one for a few hours.
For more details, we spoke with Roam Robotics CEO Tim Swift via email.
IEEE Spectrum: Why hasn't there been more of a focus on soft exoskeleton development? What are some particular challenges of making a pneumatic exoskeleton, and how have you been able to solve them?
Tim Swift, Roam Robotics CEO: In recent years, there has been an increased focus on developing soft robots and soft exoskeletons, but I agree they have yet to seem close to deployment. There are a few reasons for this.
First, the actuators most research has led to have not yet proved to be practical. Pneumatic muscles have very low stroke lengths, require large operating pressures, and create a large profile. Second, once you do get beyond the actuator itself, the power infrastructure is very poor for pneumatic systems. Our system for runners has around 800g on the ankles (including valves), but the lightest pneumatic valve that could support the power levels required by this application weighed nearly 1 kg. Using this valve would have completly negated any of our positive results. As a result, to achieve the performance we did, we developed our own valve that could replicate these capabilities for 160g. The same argument holds with compressors, and we are again developing our own approach to address these problems.
Overall, soft robots and more specifically exoskeletons have not been developed significantly because they have yet to address the major technical challenges required to make them practical in the real world.
IEEE Spectrum: Most exoskeleton companies seem to initially develop products targeting either medical rehabilitation or heavy lifting in industry. Why did you decide to make your first product for skiers?
Swift: There are two aspects to this. First, we are driven to put power devices on as many people as possible, and the ski market presents an opportunity to line up a large pool of people who are seeing their experiences limited by what their body allows with what our technology can address today. Second, I have personally lived the life of selling into the medical device and industrial markets and made the mistake once of thinking we could build a business out of these.
The problem is, despite a decade of work, no exoskeletons have made headway into these markets because our initial assumptions did not pan out. In the medicaal space the markets are more fragmented and slower than we realized. For example, the post-stroke rehabilitation market seems large with nearly 200k individuals annually in the US. However, there is not a single device that can meet the needs of these patients because of the range of needs the have. In fact, I would guess that it takes at least 5 unique devices to meet the needs of this population meaning that a stroke device is only looking to serve populations of 30k-40k patients. Customer populations of this size will, by themselves, never meet our mission of making exoskeletons at a price that a person can access it and integrate into their daily life.
How durable are these systems? How easy are they to fix if they break?
We are making a new kind of robot, so we should all be aware that there is plently to learn. That said, I think these systems will be more durable than current systems allow. Our systems are built with very few parts and our manufacturing approach allows us to be very robust to external impacts. We have logged well over 200 hours skiing with our most recent devices all over the world and not had any device failures. In addition to that, the same technology powered our early running work which saw hundreds more hours running on treadmills where we learned many of the lessons that helped us build the devices we have today.
Are there applications or situations in which rigid exoskeletons are more advantageous, or can soft exoskeletons potentially completely replace them?
There are areas where I would start with rigid exoskeletons. If you were building a device to help individuals lifting super human loads or people without any residual muscle strength, such as a paraplegic individual, I would default towards a more rigid device. In both of these applications, you are willing to sacrifice the limited mobility associated with these designs in order to complete these otherwise impossible tasks. In our case, we are focused towards activities that are almost entirely mobile. The moment you want a person to be mobile by letting them go farther, faster, or be stronger, you want a compliant lightweight machine.
What applications will you be working on next, and what kind of design changes will be required to expand beyond skiing?
The main item that I can stress right now is that winter sports is the start of what we are doing here at Roam. One thing that we are very interested in is helping people stay mobile in their communities whether they are limited by pain, weakness, injury, or just capability. When we dream of the future it involves people getting to run faster, hike farther, dunk a basketball, and stay independent in their community. These devices will require continued development in the pneumatic power infrastructure and testing with early adopters. Our plans are early, but there will be more information as we get closer.
If you’re prepared to commit to the whole exoskeleton thing right now, Roam has a very limited number of units available for pre-order for $2500-ish, with an expected ship date of January 2019. Or if you'd rather try before you buy, Roam expects that by next winter, there will be rental units at ski resorts around Tahoe, CA and Park City, UT.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.