The bionic ankle prosthesis with Elliott Rouse

Interviewee‍

Elliott Rouse is an Assistant Professor in the Mechanical Engineering Department at the University of Michigan, where he directs the Neurobionics Lab. The vision of his group is to discover the fundamental science that underlies human joint dynamics during locomotion and incorporate these discoveries in a new class of wearable robotic technologies. The Lab uses technical tools from mechanical and biomedical engineering applied to the complex challenges of human augmentation, physical medicine, rehabilitation and neuroscience. Dr. Rouse and his research have been featured at TED, on the Discovery Channel, CNN, National Public Radio, Wired Magazine UK, Business Insider, among others. He is a recipient of the NSF CAREER Award in 2019.

Transcript‍‍

‍
Elliott Rouse: Thank you. I'm really happy to be here. I appreciate the invite.

 I'm a professor at the University of Michigan and I run this research group. I would say kind of the driving force behind our work is really kind of making people's lives better. So that was maybe something that came from some of my previous experience, which we can talk about if you're interested, but I kind of, I, we used to work in professional auto racing. I loved that job and it was an amazing experience, but I felt kind of a pull to have my work impact the lives of people with disabilities and others. So I, that kind of led me to go back to graduate school and then the rest is history. Now I run this group outside of work.

 I have a wife and two great boys. So that's kind of what between work and my family that kind of occupies all the time I have.

Elisa Muñoz: That's great. And I mean, you, you said that if we wanted it to, you can share about  the experience. And that actually was my next question. Like what got you into robotics?

Elliott Rouse: So I was always pretty mechanically inclined and I kinda funneled that when I was young into being an auto mechanic, so I was not a mechanic through high school that was kind of my high school job. And then when I got into college, I ended up through a chain of events, becoming a mechanic for an auto racing team that traveled kind of around North America, did some racing in Western Europe. So that was an amazing experience. It was, I learned a ton of hands-on skills. I was there for about four years. So I kind of did full-time motor racing and I did full-time school, but school was kind of the more side part.

I was really kind of focused on auto racing at the time. And over those four years of working there, despite how exciting and interesting it was, I really felt this, this feeling of wanting to benefit people kind of more directly. And I felt a particular calling to people with disabilities. So I followed that passion into grad school. I was captivated by a medical scientist named Dr. Todd Kaycan, who had developed a surgery to kind of rewire people's nerves. And I talked him into letting me be a grad student and that kind of, at that point I switched. So I went from being a mechanic and working in auto racing to being a scientist and an engineer, fellow solving, rehabilitation challenges. And that was kinda my story for getting into.

Elisa Muñoz: Wow. At the same time. And how did you end up at Google X as a Technical advisor? That's super interesting.

Elliott Rouse: Oh yeah, that was, that was a blast. So I, that actually came through a connection I made at a conference. I was at a conference a few years ago and saw someone in Katherine Zealand and kind of spoke with her a little bit. I kind of gathered what she was doing. She was kind of doing some background research into this area and we just connected and started chatting. And I kind of learned more about what they were doing. They were investigating whether they should kind of put some resources and some effort behind this area of, of work assistive technology and rehabilitation.

So we started talking more and more and eventually the project gained some steam at X and then they wanted to form a team and actually kind of invest some resources more seriously. So I joined over there for about a year, kind of flying back and forth between Michigan to California.

And then now I kind of, that was probably from maybe 2020. I was kind of around 2019 and 2020 kind of that span. And then I left to come back to Michigan and I've been advising them ever since they're doing a great job. They just kind of passed a major milestone, which was very exciting. So it was, that was a great experience. And now it's kind of more of an advisory kind of role for me, more high level strategy. And when I was there, it was more helping Catherine kind of form the team and sketch out a research vision.

Elisa Muñoz: Wow. And they're also doing robotics in that research, or.

Elliott Rouse: I can't really talk much about what they're actually doing, but in general X desert robotics work. Yes.

Elisa Muñoz: Wow. And I mean, talking about your specific research, I saw that you had like two that you mentioned in that TEDX talk, one is a bionic new research and the other one it's related to dancing. So can you share a little bit?

Elliott Rouse: Yeah. The dancing project is hosted on my website, which I kind of created as I was looking for jobs, looking for academic jobs. So I haven't updated it much in the past few years, but my research lab website now probably serves that function. But that project was while I was a postdoctoral fellow at the biomechatronics group at the MIT media lab. So I was there for a few years after I finished my PhD. And while I was there, that was 2012 to 2014 was the Boston marathon terrorist attack. So that was maybe the last year. I think I was there and we wanted to kind of give back to the community.

We felt particularly compelled. A lot of people had lost limbs and we were kind of a lab that specializes in advanced prestige seas for people with amputation. So we kind of wanted to get in there and help. So he threw in kind of a series of connections, found Adrianne Haslet-Davis,who was formerly a professional dancer before she lost her leg below the knee and the terrorist attack. And we kind of wanted to use our sort of viral robotic tools to give some of that functionality back, at least to the best extent that we could. And we asked what we did. So we developed a controller, we analyzed the way professional dancers move and how their joints respond and interact with the ground.

We programmed a robotic ankle prosthesis to kind of behave that way. And she returned to the dance floor. We demoed that at the Ted conference, which was fantastic. It was a beautiful and wonderful project to work on.

Elisa Muñoz: It was definitely a great one. I mean, I saw her face and she was nothing but grateful and, you know, everyone was really moved about it. So I think that applying robotics into the human part of it, I think it's also great. And talking about the technical side, how hard it was to build a robotic ankle and knee and how does it work?

Elliott Rouse: Yeah. So maybe for that, I'll talk about some of our newer projects we have, which is the development of an open source robotic leg. That's been kind of going on for me for the past few years. The impetus for that project was really that the control of these systems is their major barrier, getting them to act intelligently so that they don't let a person fall to the floor or fall out of stairs. That's very, very challenging.

It's kind of a different game than some of the upper limb prostheses that don't have that constraint of holding the body up. So we wanted to provide a hybrid platform to researchers around the world who could study these control challenges and answer them so that we could get this technology out and into people's lives. So we developed mechanical test beds. So basically have a robotic knee ankle press thesis. That's been designed from the ground up to be easily manufactured, assembled, and controlled. And we have a pretty extensive website. That's open-source leg.com and it includes all the information to build on yourself. So all the part files, drawings, the bill of materials, there's assembly instructions, detailed videos, teaching somebody how to assemble and disassemble it there's control files.

So software you can download and videos demonstrating how the software works. So it's actually, I got to develop the website. It was probably the biggest undertaking of that entire project, but that was a fantastic experience now. So now we developed one leg, one design, and it did well. And then we have now developed a second design based on some things that have been developed in the meantime. And our second design is now released on the website. It's also being commercialized by a company called homo tech that works in our space. So that instead of, for people who don't want to build it themselves, which is probably a fair number of people.

So they can now just buy it from this company called Motech, but this robotic leg is kind of intended for researchers, at least at this point, like there's no, there's no pathway for that to get out of the lab and into the lives of patients. It's more of a research platform to accelerate the research that's going on in the control side.

Elisa Muñoz: Well, this is really interesting. What are the greatest questions we can start to answer with these open source legs?

Elliott Rouse: The greatest questions are, how does the leg know what to do? Someone's walking. It has to provide what are sort of cyclic patterns of motion and effort. And it has to know, are you walking if it's so common to break up, this problem is to say, let's set, let's first decide what activity someone's doing. And then once we know that we have an activity specific control strategy that we would use. So It would always be trying to figure out if someone is walking, are they standing? Are they going to go upstairs? Are they going up an incline or decline? So it's kind of always running that intelligence behind the scenes.

The greatest advancements would be one to solve that, to solve that challenge so that we would like right now we have systems that are 99%, even greater 99 point something accurate, but that still means one step at a hundred to one seven, a thousand. There's a mistake which could be catastrophic. So we need this kind of problem to solve that control problem, but then the higher or maybe bigger picture. I think the greatest improvements will come in the future and we can directly link these systems to the human nervous system right now. That's not possible because information coming from the nervous system is really uncertain. It's really unclear.

So a lot of intelligence goes in to figure out whether those commands being centered are accurate, or whether we know what they're actually saying or decoding. So for the lower limb, we don't use neural signals at all because it's just too uncertain. So all the information is onboard. All the sensors are onboard, the press thesis. It's really walking, like operating totally in parallel to the human. And it has to perfectly guess what the humans are doing all the time. And that's just a very, very challenging problem to solve.

Elisa Muñoz: Wow. It does sound like a very challenging problem to solve. And I mean, does it need batteries? Do you need to charge it somehow? Or how does that work?

Elliott Rouse: Yeah, so the, I mean, in basic terms, the motor, I mean the leg is a motor, a transmission and a battery. So that's kind of like how it operates. It's just an electric motor with a transmission and it's packaged in a way that makes it look like a leg. But yeah, the second gen version has an off the shelf battery that just sort of slides in. So charters in a receptacle, it has what's called onboard battery management system, which is safer. We want it to get away. We used to use just a lithium polymer, pouch batteries, which are a little riskier. So now we have an off the shelf battery that is used. It's integrated.

It's a little bit more user-friendly I would say.

‍

‍

Elisa Muñoz: And does it have to do anything with drones? I heard something about it, but I didn't quite understand. So how does it work? 

Elliott Rouse: So that, so when we were designing the open-source leg, one of the decisions we have to make is what motor do you use? And recently the drone industry has created a motor that looks a little bit different than a regular motor. These motors are wide and thin because drones can't have transmissions. So they need to create the torque. They need not by using a transmission, but by increasing the diameter of the motor. So instead they have these large diameter kind of thin motors, and we leveraged that to reduce the transmission ratio in the design. So now we can not have as high of a transmission ratio, higher transmission ratios come with challenges, reduces efficiency, it's harder to package. So we basically leveraged this new style of motor, these motors from the joined industry, for some advantages that open doors on the design side. And that worked, that worked very well.

And now I'd say it's more common in robotics to be using these notorious. They've, they've made a big impact.

Elisa Muñoz: What was the biggest technical challenge you managed to solve when it came to implanting the drone into the open source leg?

Elliott Rouse:  I guess there's one, like one challenge specifically around these motors is that when we first use these, so when you make the motors really large and they tend to produce a lot of torque, they have lots of what's called POL pairs. So they have a technical attribute to them that can be challenging. On the computational side, they're run by a small computer called a drive. And so when motors had these maypole pairs, we couldn't get a drive to use it. You know, the way drones are controlled is different from the way robotic systems are controlled.

Elisa Muñoz:  Have you found any critical challenges when it goes through the procurement process?

Elliott Rouse:  Yeah, so we tried one of the constraints that we kind of put on ourselves when developing this leg is we wanted to minimize the number of vendors or suppliers that are used. So we actually worked hard to consolidate all the parts that are ordered from kind of like three or four companies.

So in that sense that wasn't actually the hard part was finding parts that aligned with that philosophy. But once we did that, the procurement process was pretty straightforward. We do most of our machines with machine jobs in China, and there have been some challenges getting these tariffs, getting things back and forth, but overall we are very happy with the quality we get in the parts we get from them. So yeah, I'd say in general, actually maybe one thing is for the open-source leg, we, we partner with this company defy, who I mentioned, who makes these controllers that run on the motors that makes them work okay.

The five can't sell or don't sell certain countries, or they have a kind of a limited set of countries, they're defense contractors. So they have a limited set of countries that they can interact with. So now let's say we had somebody interested in working in open source, who is not from an allowed country, then they can't get access to the electrical hardware that they need. So what we did was figure out a way to have it work with the Russia splinters drives that come with it. We kind of gave him a second option so that we kind of solved that problem a different way. And now we've developed kind of an open source library.

Someone can just download on the internet and it runs with these others, a totally separate set of kind of computing and drive hardware. So that was like a, that's kind of a procurement issue and also speaks to some of the challenges that we've had to overcome. And some of the exciting parts about people around the world being interested. I work with a group in Armenia who just got a grant to develop an opensource leg program in Armenia and like.

Elisa Muñoz: Last but not least, do you have any advice for future engineers or people starting on this path?

Elliott Rouse: I would say the thing that I see. I talked to a lot of students that come in and they're kind of trying to figure out what to do, what they should do with their life. And one thing that I see as a common or consistent theme or trend is that they're stressed and that they're really concerned about. They gotta find kind of the one, the one thing they're meant to do or that, you know, though it's a right answer of what their next step is and they're kind of stressed about it. And I implore them. Like, I don't think that's the right way to think of it. You know, you don't want to think there isn't just one thing. There's many things.

And if you kind of identify like what you like is that specific type of tools that you like to use, or a type of problem you like to solve that opens up many options, not usually just one. And I kind of try to get them to think about it that way and, and not put so much pressure on themselves. Like if I was working at a different place, I would, I would probably love it. Like there are probably things that I would find about that job, assuming it's something reasonably around engineering that I would really enjoy. And I would focus on that and I would find, hopefully find some level of like interest or fulfillment, but yeah, so I kind of talk them through that concept. 

Elisa Muñoz: Wow, this is really impactful! Thank you so much Elliott for being here and sharing your experiences with us.

Elliott Rouse: Thank you for having me. This is a blast.