Full transcript: Vytas SunSpiral on London Real
0:00 Brian Rose: This is London Real. I’m Brian Rose from London Real. I’m in beautiful Lausanne, Switzerland, to attend a lecture by Dr. Vytas SunSpiral. He’s a robotics engineer at NASA’s Ames Research Center. What he does is he likes to use human movement and human motion to teach him about robots that he designs to build and go into space and land on Mars.
0:35 So we’re going to go to the EPFL, which is the Eco-Polytechnic Federal d’Arson, which is the local technical university, and we’re going to go watch him now. So let’s go check him out.
0:56 Vytas SunSpiral: To really understand my motivation here, you know, I’m really interested in understanding how we think, how we feel, our intelligence, and how we do everything that we do. And to do that, I think it makes sense to start by understanding how you move. And that’s from the basic assumption that the only reason to have a brain is to control motion.
1:36 Original. Now, we appreciate art and all sorts of other wonderful things, but nature starts by the brain is a control system for motion, and once the basics of that are evolved, it gets used again and again in more complex ways. And so to understand the real fundamental motion is to get at the rest of the understanding of the brain. And the way I’m going to talk about this is, you know, if you want to understand the control of motion and how we move, it’s really important to start by understanding, you know, our physical system.
2:11 What is it that you’re moving? What is the structure? And there are some emerging ideas on our bodies that are very different from our natural assumptions of how we work, and so I’m going to talk about that for a little while. This is really looking at the body and how the connective tissue and the fascia makes a tension at work that is very similar to a tensegrity structure, and I will talk about tensegrity structures after that, and where we are, they’re again somewhat counter-intuitive.
2:50 And then I’ll sort of tie that together into this idea of the body as a tension network. Then I can share a little bit about my current research, building some robots and a physics-based simulator to explore these ideas. And then finally I’ll get into where I’m going and I’m sort of on the edge of really diving into the controls research. Animals that are moving change their relationship to gravity all the time. We don’t just stand upright like this constantly.
3:23 We do handstands. We can roll. You look at any animal moving in nature, they go up and down slopes, they fall over, and they have to deal with this. If you look at a lot of the classic biomechanical analysis of our bodies, there’s this assumption that we are upright all the time. So you see lots of assumptions that say, okay, well, if you look at the vertebrae, the small ones at the top, it’s like a pillar, it carries 9% of the load, and the one at the bottom carries 60% of the load, which makes sense if you always stand upright.
3:59 But since we can do handstands, I can do handstands, right? But collision avoidance is not perfectly possible in the real world. There are other actors out there, other things happening. So you have to assume that there will be collisions. The newest version is four meters tall when it walks, and I spent time working with it, designing walking algorithms.
4:21 But when you see this skeleton, maybe you had this in your classroom in biology, it’s misleading. You know, they had to add aftermarket, they had to add pins and joints everywhere to hold it together. In the real skeleton, there’s nothing that holds it together rigidly. It’s all soft tissue. And this next picture is from an actual cadaver from a Body Worlds exhibit.
4:44 And here you start seeing the fascia as a very large just structure that connects all throughout the whole region. And the muscles attach to it in different locations and the bones are connected to it. The common view is that, oh, it’s vertebrae disc, vertebrae disc, and the weight just passes through it. But if that’s the case, then you wonder, well, why do sometimes we get bulging discs and other times not? What’s the difference?
5:11 And the difference is that under normal conditions, force should not pass compressively from the vertebrae to the disc to the vertebrae. The force should pass through the tension network and not compressively through those discs. You’ve all heard of a chicken with its head cut off, and it’s a funny joke, but it’s real. And what happens when you cut the head of the chicken off is it runs around like crazy. So it’s worth thinking about.
5:40 Where the chicken is going is not a random. It doesn’t have intention. It doesn’t have a goal-seeking behavior. It doesn’t know where in the world it’s headed. But to run around is a highly coordinated action.
5:55 This is the example of the cat that has had its brain stem severed but kept alive. And you’ll see that as it walks on this treadmill, It has entirely different gates, so it goes from walking to trotting to running, and again, no central computation. This is where I get to brag about the work you guys do here, and obviously this is work from Alex Ladd, where this robot is moving with implementation all from CPGs, from central pattern generators, and it’s capable of, you know, a wide variety of behavior with very simple controls.
6:40 Thank you very much.
6:42 Brian Rose: Hi, it’s Brian Rhodes from London Real, and I’m here with Dr. Vita Spiral, who’s a robotic engineer with NASA at the Ames Research Center, and I just watched him give a talk at the EPFL, which is the Echol Polytechnic Federal Lausanne, and could you tell me what you were talking about and how you find humans inspire your robotic work.
7:09 Vytas SunSpiral: Sure, Brian. So the core part of my presentation is this interesting idea that if you want to understand how we think, how we humans think and appreciate art and do all these things that we love to do with our brains, we should start by understanding how we move.
7:31 And this is, it may sound strange, but it starts because The only reason to have a brain is to control motion. If you think about that from an evolutionary perspective, your neural system, your neurons, are controllers. They control your motors. They integrate information from your sensors. And so that’s where the first steps of evolution in terms of developing neurons and brains comes from is to control your motion in the world.
7:59 And so if you can really drill down and understand the basics of how we move, then you can start finding those fundamental building blocks that impact our whole control system, our brains, our neurology, our physiology, and nature reuses those same tricks again and again. And so you’ll find those same core qualities all throughout the higher processes of our brains, even getting to the point of understanding how we relate to each other, how we talk, how we think, and it’s all going to sort of start from those same fundamental basics.
8:31 Brian Rose: So I was watching your lecture today, and I noticed that you relate first to the human beings and the living creatures in order to go to robotics, where I think a lot of traditional methods is the engineers start at kind of the civil engineering, structural engineering they’re taught, and then go to try to mimic biology.
8:50 Could you kind of explain your ideas there and how you approach it?
8:54 Vytas SunSpiral: Sure. Yeah, and it’s really kind of a question of how do you build a machine versus doing fundamental science about us. That’s really the difference. And a lot of folks, they’re doing a really great job building machines to do tasks that we want done. And that’s useful, but in some respects that will only ever get you so far because you’re always building up on technologies you already understand and pushing those limits. But to get the really disruptive, game-changing new ideas, you kind of have to attack a problem from a different angle. And a lot of the Other things that we’ve learned how to do with robotics really were successful in environments that were really well understood or controlled, factories, laboratories, you know, reasonably controlled settings.
9:43 And I’ve worked in all those sorts of environments, but as you try to take robots into more and more complex settings, like out into the natural world and dealing where there’s other things is moving around, other people, other robots, you know, other animals, and you don’t have nice flat floors, you know, and all these, the world gets more and more complex. You start finding that the traditional approaches really run into limitations. They get really, you know, they don’t necessarily succeed at doing the things that we find very easy.
10:13 And so that really is then the reason to start changing the question and saying, well, Why don’t we really try to understand what we do and get in there, partly because it’s just darn interesting. I mean, at the end of the day, I think humans are the most interesting thing out there. We’re far more interesting than any piece of technology. And so let’s study that. Let’s learn how we work.
10:37 And then use the robots, use the implementation as the mirror by trying to build robots. to then come to an understanding of what’s hard about this problem? What’s it take to walk? What’s the quality of the problem that needs to be solved here? It’s so intuitive for us that we don’t even think about it most of the time.
10:59 So building the robots sort of lets us understand what’s even the nature of the problem and then we can look back at our own biology, our own physiology and neurology start seeing, oh, maybe, maybe that’s part of the answer, you know, and pull inspiration and ideas from a study of ourselves.
11:17 Brian Rose: Wow. Now, when you work at NASA, are you trying to build a lot of these robotics that are then going to go land on Mars someday, or what is, what are some of your end goals when you write these robots?
11:29 Vytas SunSpiral: Well, you know, first of all, my work at NASA, I should say, is I’m involved in quite a number of projects, including other robots that are, you know, free-flying robots in International Space Station and robotic systems for planetary exploration. We do a lot of different things there and it’s good to sort of have a hand on a lot of different projects. My particular interest in these tensegrity robots and looking at the body as a tension network really comes from the question of how do you build robots that are going to really manipulate and touch and interact with the world and walk in the world in ways that are going to be very energy efficient and that’s one of the challenges for NASA is that our robots have to be lightweight because we launch them into space and and we have very limited amounts of power right because if you’re gonna send a big battery up or a solar panel or whatever you’re gonna have limited power not like here where you can just plug it into an electrical grid and have as much as you need so looking at new ways to build robots that are very lightweight that can be compact and then deployed so that you know you can launch them into a small size and then have them get bigger once they’re in space. These are qualities that you can also see and I didn’t talk about it during the lecture but these are qualities that are part of the Tinsagrady robotic concepts that I’ve been discussing and because they can be such lightweight structures it
12:52 Takes less energy to move them and some of the ideas I talked about in the presentation were how we can use things like oscillation, how we can use, you know, the vibratory nature of these robots to generate motion. So you can maybe excite that oscillatory nature of the structure to get it to move and that can become a really energy efficient way to move. And so those are There’s some of the reasons why NASA’s interested in this so that we can build efficient robots to do, at this point, whatever we may want to do, exploration, manipulation, deployment of large structures like space antennas and whatnot.
13:41 At this point, I’m really looking at the fundamental science of it, and I’m not pushing too hard on the specific applications. I have some ideas of some nice projects to deal with it, but sometimes you just have to really start with the core science and then worry about the employment later.
13:56 Brian Rose: Okay. Do you watch the Olympics or watch, you know, the ultimate fighting championship or do you watch humans do different movements and then kind of get inspired or get ideas from that?
14:06 Vytas SunSpiral: Absolutely. You know, a lot of the work I’m doing is really me integrating both my professional intellectual career of robotics with my own personal exploration. I love to do martial arts and yoga and dance and all sorts of sort of trained and very physical activities and when you really take your your own physical activity to you know a very advanced level you start getting insights into how you move you can only be a great martial artist if you really understand how to generate speed and force and power with your body and you can only really do you know you can only advance in yoga so far if you really understand some of the alignment and some of the subtle details of how you can rotate parts of your body to expand your range of motion and how the chain of connection in your body works so that even when you’re trying to afford bend what you’re doing with your feet changes how you move you know and how that’s all connected and so it’s by combining these different aspects of my life right the my own personal experience of motion and sports, you know, martial arts and yoga with the analytics of robotics that’s now kind of coming out in this research area that I’ve been discussing.
15:26 Brian Rose: You know, I did some shadow boxing after your lecture and I was thinking to myself, you know, is it the fascia, that’s what you call it, the fibers that were pulling my leg around to throw like a tie kick and you know I was thinking how you were saying that the bones weren’t actually touching each other in a body so you’re not necessarily having everything in a compressive force. And when I was thinking I had to be on my toes when I throw a punch, I was just thinking in so many degrees of freedom that the human body has that I don’t really think about. And I think having that knowledge might make me actually practice martial arts differently.
15:59 Vytas SunSpiral: It really could. I mean, I certainly find it useful for me to understand how we move, and it gives me a better model to move myself. And that’s the interesting thing about it. We don’t control motion consciously by saying, okay, motor number 37, turn on, okay, motor 38, go, you know, it’s not like that. We, due to the complexity of our body, we use very abstract imagery to control our own motion.
16:32 We use very abstract visions of like, well, I want to move, you know, in a spiral this way, you know, or I want to, um, I think about motion because I’m just sitting here, I can just think about, you know, the energy flowing down my back and that causes my shoulder blades to go back and that changes my whole structure here and the power in my shoulders and it can set me up for a different type of strike than, uh, I would have just slumped like that, you know. So by having a different understanding of how your body works, you will also then, uh, can have a different visualization. By changing your visualizations you can make more effective motion just by how you see it
17:13 Brian Rose: that’s interesting I mean how do you apply that to a robotics perspective I mean I think you might my jiu-jitsu instructor always said posture Brian it’s posture everything whoever controls the other guys posture wins I mean how do you relate something like posture which is very abstract yeah into a robotic framework
17:33 Vytas SunSpiral: right well so the reason why he’s saying postures is what helps you win is because if you have a good awareness of the alignment in your body, you can really maximize your force production and maximize your stability.
17:46 Brian Rose: Final question, you have the coolest surname I’ve ever heard, it’s Sun Spiral. Yeah. How did you get that?
17:53 Vytas SunSpiral: Well, I got that because when my lovely wife Claire and I decided to get married, you know, we’re both awesome people. She’s an amazing human being. I have enormous
18:03 Respect for it. And so we felt that giving up one person’s name for the other, as this tradition didn’t seem right, you know, were equals. And so what are we gonna do? And you know, Haifa Nation, putting our two last names together with the Haifa, it’s a one-generation hack that’s non-extensible. You can only do that once, maybe twice, and it gets out of control, right? So we decided that the best way to do this was to make a new name together that was both of us equally.
18:30 And I used to do a lot of medieval martial arts, and so I had always used the sun as a symbol on my shield, as my token of who I am. Coming from my old, I’m half Lithuanian, and that’s a symbol that’s very popular in the Lithuanian mythology. And Claire has always loved this spiral as part of nature and its sort of mathematical beauty. It had taken spiral as part of her middle name legally, right around the time I met her. name to be Claire Spiral by a Rose and so then when we got married we took these different symbols and you know played around with it for a while and eventually came up with Sun Spiral as our new family name.
19:10 Brian Rose: Wow that’s a cool name and a cool sister. Sorry cool wife who was also my sister so I have the coolest brother-in-law ever. Well I’d like to invite you to London some time to come on the show so we can get a cool hour with you on London Real but Thanks so much for the time. Thanks for letting this film film you today.
19:30 Vytas SunSpiral: Sure thing and I appreciate it, right?
19:32 Brian Rose: All right, take care and we’ll see you next time until then it’s about the journey
