Mushroom Robots with Organic Robotics Lab
Alisa Nappa
Could mycelium power robots to solve large scale problems? Today we sit down with Rob Shepherd & Anand Mishra of Organic Robotics Lab at Cornell University to talk about their new mycelium powered robots and all the fun things they can do. Tune and and shroom in.
Sign up for our podcast giveaway here. Our next winner will be selected on October 23, 2024 and contacted via email.
TRANSCRIPT
Unknown Speaker 0:00 Alex, Lera 0:12 welcome, welcome. You are listening to the mushroom revival podcast. I'm your host, Alex Dorr, and we are absolutely obsessed with the wonderful, wacky world of mushrooms and fungi. We bring on guests and experts from all around the globe to geek out with us and go down this mysterious rabbit hole to try to figure out what the heck is going on with these mysterious fungal friends of ours. And today, we have some researchers joining us to talk about a fungus controlled robot, which is pretty, insane, pretty science fiction, and I'm very excited to talk about it. So how you guys doing Speaker 1 0:47 great? Alex, thanks for having us on. It's amazing to see that you have a podcast dedicated to fungus. It's awesome. Lera 0:55 Anand, how you doing? Speaker 2 0:56 Yeah, I'm doing great. And I'm really excited about, you know, talk about how this fungus robot works. Lera 1:03 So we only have 20 minutes. This is kind of a shorter episode, but so let's just dive in. If you want to give your introduction, introduction of the project, how? What was the inspiration about building these fungus controlled robots? What's the elevator pitch of what you guys are doing? Unknown Speaker 1:21 Yeah, so Lera 1:23 we Speaker 1 1:25 my, personally, my lab. We were called the organic robotics laboratory at Cornell University. What that meant was we were trying to make robots out of polymer chemistry, carbon based chemistry, but it didn't mean organisms or anything like that. This we started about 12 years ago, and so we're using synthetic chemistry to make robots more lifelike. And along the way, biohybrid robots became more possible. More and more people were doing them, but they were doing with usually mammalian tissue, and that is quite expensive and difficult to work with, and I just didn't want to devote the time to get good at it. At the same time, I had just found an area of woods that I was exploring, and I was finding mushrooms, morels, things like that, and tried to culture my own mushrooms. I got some bolts and put them in some logs. It didn't work. But either way, it seemed like that's a kingdom that wasn't being exploited, that could be with a lower barrier to entry, to work on biohybrid systems and at the same time, Anand was in my groove, and is just such a I don't know he's brilliant. Cares about the world and nature and everything's interconnected. Listening to him, you start getting very spiritual, even though I'm not. And I think he had his own independent motivations for working on this, and we just kind of, I'll let him explain that side of it, but for me, I wanted to work on biohybrid systems without having to deal with the financial and learning curve mammalian tissues. Speaker 2 3:17 Yes, so thank you. Thank you, Rob and yeah, so I'm, I'm a Research Associate in Rob's group at Cornell University. So I think the motivation for me to work on that was actually started with, you know, like, some sort of, I watched a BBC documentary they were talking about, you know, how mycorrhizal fungi can, know, control the below ground economy, like, they can come like, you know, they can somehow share resources with trees. And then, and then Rob was already interested on that, and we were just like, and I started my job here, like Cornell in, like, couple of years ago, and then that's where we were talking about this idea. And Rob says, Let's build a robot. And that's why I think that was a turning point to, you know, think about how fungus can be, you know, fungus system could be a part of, you know, robotic model for robotic systems, although the when we started, we did not, we have a very little knowledge about fungi. But, you know, over the year, we grew actually like fungi. We also grew about, we build expertise. And, you know, learning about, you know, the what fungi can do. And there are a couple of things that you know we more scientifically, that we can talk about is the fungi are very robust organism that they can grow in, like very extreme conditions, like Arctic or saline conditions, and that makes us our approach to use them for robotics. Makes a more bold step that okay, we can probably if we utilize this organism as a part of robot, it would be really cool system. Lera 4:48 So I know fungi, and I know they're extremely unpredictable, and they kind of break every single rule in the book, and every time you try to put them in the box. They, they, they find a way to get out of it. And so, you know, but I don't know robots and and robots seem extremely predictable, and you can program it to be to do exactly what you wanted to do. So for me, they seem like a complete juxtaposition. But, but, yeah, but, like, Please enlighten me. Like, why would someone want to power a robot using mammalian cells or plant cells, or what you guys are doing using fungal cells? Speaker 1 5:28 So nature builds from the bottom up, and we build from the top down. So if you're when you're building from the bottom up, at the for the protein level, amino acid level, you can get incredibly intricate structures that can do things we will never be able to do using artificial techniques. So you get these complex systems that can do things like autocatalytic upregulation of chemistry. Biology is very sensitive to other bio biology and chemicals, either to get food or to avoid attacks. So you can use nature that way, but it's at a higher level. You asked, Why would people you do bio hybrids in any sense so at movement, signal processing and computation, I think would be the three reasons. So sensing, computing and actuation, nature is better at all of those things than we are. The question is, can we harness any of those things artificially? And primarily mammalian biohybrids are used for actuation purposes. Muscle is a very good actuator compared to human made ones, not in some cases, human made ones are better. In most cases, for general utility, natural muscle is better for people who use plants as well in biohybrid systems for sensing, even for growth, slow, very slow actuation. Sometimes you can have rapid responses, but the fungal kingdom was not explored for biohybrid systems very much at all, not integrated into robots. And they're very sensitive to chemistry and light and and things like that. So that's, that's the what we wanted to use them for, to be honest, in our in our paper, we didn't really use them for what they'd be best at. We used it for light based signal sensing, sensing UV light, things like that. But we have semiconductors that are very good at sensing light, so we didn't use it so artificially. It still would do better to do that, but the point was to figure out, and Anna will talk a lot more about the details of it, how to integrate it into a robot, and then, once it's in the robot, we can start using for what it is best at biological and chemical detection. So anyway, but which, which we won't be able to do artificially in a multi input, multi output way, like we can make very tuned, very specific chemical sensors artificially, but we can't make ones that are going to sense the multitude of chemical inputs of the environment and make some kind of sense of it that we can interpret as information to respond to. Lera 8:47 So Anand, let me ask you this. You know, we use dogs all the time to sense, you know, cancer, and they can smell cancer in people, which is pretty wild, or, you know, explosives or drugs. And you know, that's our partner organism first, and its superpower is sense of smell. And, you know, like Rob was just saying, fungi superpower is sensing, you know, chemicals. So what would be the practical use case of these fungal robots in sensing biochemicals, Speaker 2 9:24 yeah. So, so I've me so, as Rob mentioned that in the paper, we talked about, we have used light, you know, and these lights are actually very well, like, you know. It is known that the fungi, like, you know, doesn't like UV light, for example, one of the use cases, use and but if you look at me more in terms of, you know, the properties from this group, we look down at the soil, right? And if you see that in a soil, there are lot of microbiome make lot of chemical compositions. That is. Statics is, and that's for something that we wanted to, you know, look into. We want to see that okay, can, because there are a few questions we don't know even like, you know, we we have, we haven't even solved yet. So we are working on it. And we wanted to look into that can fungi, can sense, like, different chemical compositions, we believe, like, we think, yes, they can, because they live in that those kind of conditions we have to probably check it, test it, train it, and try to, you know, like, know, straddle data, but definitely, mostly the chemical inputs or biological inputs that we think that, that that's that we can actually read out, from, from, from fungi. And as Rob was also mentioning, and I think the reason is because you can make a sensor that can, in sense, one input, or you can't have a sensor that can do like, you know, just like a multitude of inputs, that is not possible through synthetically or synthetic chemistry. So that's why. So in more specifically, we probably can sense, like maybe some chemicals, some gasses like CO two, or we can sense like some sort of some nutrients like phosphorus, or some kind of other kind of ions. That's what something we are interested in, and we hope we can do it. Lera 11:16 And I read in your paper that one potential use case would be to send soil chemistry and row crops and decide when to add more fertilizer. How would, how would this practically work? So you have, you have a some fungal cells and a robot, and then what you know like for for someone who does not have a background in robotics like I don't know if there's an easy way to describe how this might work, but I'm just curious how that would go down. Yeah. Also, Speaker 1 11:51 Alex, I managed to extend. I moved my meeting 15 minutes back, so there's another 15 minutes. Perfect. Great. So, okay, so robots are to do. Robotics is difficult in terms of like control and all the other stuff, but to understand them, it's quite simple. They sense things, they compute things, and they respond by actuation. So at a very basic level we would have we can also talk about why we chose mycelia over fruiting bodies, for example. But let's say we have the hyphae bundles distributed across little rods that can stick into the ground. We can push the rods into the ground allow them to respond to the chemicals in the environment, or parasites. What would be parasites to plants, but maybe not to the fungus, but either way, there would be a response, we think, and then cause the response to be injection of fertilizer as an example, or of a nematode pesticide or something like that. I mean, it's very simple, one tube in of the mycelia, another tube out of the chemical, and then a little computation, decision made in between. So Lera 13:12 it it's kind of like a more advanced version of a canary in a coal mine. You know. Unknown Speaker 13:19 Yeah, that's right, yeah. Lera 13:20 And what are some of the drawbacks of using this, this biohybrid method, I Speaker 1 13:28 think that's a chance for you to talk about all the stuff you had to do to make it work on it Speaker 2 13:33 wasn't easy, yeah. So I think the one part is the life, you know, it dies. And let's say When, when, whenever you connect. And then the, I think the one thing that we didn't mention that is our approach is basically how we interfacing robot and this living system through electrical interface. So we are reading electrical signal. There could be other ways as well, like you can try optical or other interfaces, but in this, and the thing is that that interface is a kind of a challenging job. You have to make sure that the mycelia has a robust and strong connection with this, so that you can read out those signals that we are looking so I would say, I think the like, you know, making them, for example, you know, usually in our lab that we grow these mycelia, and they can work like they can grow for months, but we haven't tried something like, let's say, in a very like, more unknown conditions we're seeing like, how difficult it is going to be. So I think, I think the maintaining them, you know, growing in a more robustly, and making those into, like, elliptical interface, more stable connections, so that we can read all those signals without noise or with minimum, like signal to noise ratio, that is, I would say, the biggest challenge. Lera 14:48 Yeah, it is very similar in both micro remediation, so using fungi to clean up toxic waste in the environment, in the lab and on paper, it works great. But then when you. Put it in the field, it's a living organism, so if the conditions are not right, it fails. And the same thing with Myco pesticides, you know, using fungi to attack insects in the lab, it works 99.9% effectively. But once you put out in the field, you know, if the farmer sprayed fungicides, it ceases to work, right, because there's chemicals already in the environment that are killing the fungi. So, yeah, I mean, I can imagine all the different things. Or if you pick a species, right, and it works really well in a certain environment, and then you move it to maybe a desert, or something like that, doesn't have the capabilities to survive as well, even though fungi as a whole are pretty robust and they can adapt really well, they're still living organisms at the end of the day. And I'm curious, have you worked with many different species. Are you currently focusing on one and what? What was your, what was your thought process of picking certain species? So just Speaker 1 16:11 to go back to what you said, yeah, it getting them to work in the field is why on its work is a big deal because it is in a machine that is roaming around the hallways, and it works like wow. And I think the how we got that got to that point before other people is at least Anand is very thoughtful, and he wants to learn about how the mushrooms work and what the hyphy signaling means, and all that other stuff. And I was quite adamant that let's just make this work and then get it to roam around, and all the science can come after that. So the choice of the mushroom initially, he just picked something that we can get off of Amazon and and that that we wouldn't have to have any environmental health and safety issues around we can just start working with it. But he started with pink oyster mushrooms, I think, right on. And, yeah, Speaker 2 17:18 yeah. So, so I think I ordered these because, as we mentioned, we then we started this project. We have, we haven't worked in this area, in this field, so just the first thing was to buy a pack, you know, just growing kit from from Amazon. And I order, like, pink oyster and king oyster. And over time, you know, by working with different, you know, because, as I mentioned, the electrical interface is a very important like problem to find out this key moisture is very filamentous and and whenever you connect, you know, with our electro system and our scaffolding that we build, we can start reading out the signal. And that's became our choice. I tried. There is a one fungi called Neurospora crassa, which is, like, you know, model organism in this field. Tried, they're not filamentous fungi, so I did not get a very good outcome as I was expecting. But this, this was the like, you know, that's what our story became, and, and I think also this is a kind of a first organism in terms of, like, king oyster, to study the electrical signal. So that's Speaker 1 18:23 maybe you can talk about why we didn't use fruiting bodies. Speaker 2 18:28 Oh, yeah. So do you want to do you want to talk about, please? Okay, Speaker 1 18:32 sure, because we want these robots to be persistent over as long a period as possible. We didn't want to rely on the growth and death of, I don't know if you call it death, but actually, Alex, you probably know much. Yeah. I Lera 18:48 mean, it decomposes, yeah, for sure. I mean, it's like, yeah, it's, it's not the main growing organism, right? It, it, it pops up and then it decays pretty quickly. I mean, with some exceptions, like some conchs grow a lot slower, but most, most freeing bodies or mushrooms pop up. But then, honestly, Speaker 1 19:05 I feel like we can learn more from you, Alex than you can. Lera 19:09 Oh no, I don't know anything about what you're doing. I'm having you on Speaker 2 19:15 just to add that, like my Celia is a kind of vegetative part, right? That's what it everything it controls so and I think, I think I started working on using food in body. And then there was one of the meeting I had with Rob, which I remember that I said, No, no, we should focus on my cilia. Stop doing food body. But I think it makes sense, because, you know, the they were growing, you connect electrode, but after a couple of weeks, you see that they are just, you know, started drying, and you can't control them, and they just just die. But in the case of my Celia, when you grow them in in a just electrical, like no player, you can keep them for like, couple of months without a problem. So that that was our way by, like, our choice, to Speaker 1 19:54 an advantage of a fruiting body, though, is the action potentials were fire in them. Yeah. Go ahead. Oh, it's okay. Well, Anan didn't say didn't say this, but the action potentials he was measuring from the mycelia were 10s of microvolts, which very small to measure. And that's why in one of the videos, there's this Faraday cage over it to prevent EMI electromagnetic interference, right on 10s of microvolts for that that he was measuring, and to prevent the motion of it from jostling the electrodes that he had in there. But what were you getting for the fruiting bodies? It was hundreds of microvolts, maybe millivolts, I think. Yeah, Speaker 2 20:39 it was in range of milli board, if I remember that, like, it goes to, so sorry, at least you can say, like, like, 10 to 100 times, because signal you get. And I think what the reason is because you have a better electrode, like, more more, like, you know, the contacts are much larger and right, higher synthesis exposed, Lera 20:57 yeah, king oyster mushrooms are pretty beefy. Speaker 2 21:02 Uh, and I think that that was the that was the point, but, but then, you know, you keep the electrodes there and keep it for a while, but then you see that the contacts are losing. The signals were unstable. That's another problem. The signals were, you can see there is a like, you know, very huge, right? Because it's a couple with AC and DC board. So it creates a problem. So, and that's what the the mycelia, with the scaffold that we made you, you like, know, you have much stable connection, so you you don't have these kind of more big, like a moving signal, so you can have, like, a fixed phase, and so you have less DC part to it. Alex, if you Speaker 1 21:42 were trying to pick my ceiling, a species that had a very dense mycelium network, filamentous network that was relatively that was safe to work with. What would you have chosen? Lera 22:02 I mean, you brought up ropey mycelium, and the first thing that came to mind was King straferia, or wine cap. However, they hate sterile culture. They love it dirty. So, I mean, trying to work sterile in the lab, it's the mycelium just doesn't do well. And then the second year you're kind of growing. It pretty dirty it, it grows pretty rapidly, but, but that Mycelium is incredibly ropey. Okay, another one i is, is our malaria that comes to mind. They make these melanized, super ropey mycelium, I don't know. Again, cultivation might be hard, but the melanized sheets come to mind. They make these kind of melanized sheets over the mycelium. And could have something to do with electrical conductivity, is my guess. But cultivation, I have no idea how you could do that and input it into a robot. Those things come to mind immediately. But, you know, I'm sure once I once my coffee kicks in, some others will will come in again. But, yeah, I don't know. I mean, it's so interesting to see this world. Most people kind of work with these electrical impulses to make Have you seen the mushroom music, where people kind of put electrodes to mushrooms or mycelium, and they make music. We brought in another guest that was using the same concept to input the data into VR, and based on their I think is EKG or something, something like that. They're, yeah, they're their own data. They basically control the VR as whatever they're feeling changes the VR reality, and then also the mushroom input, and other people using mycelium to create new batteries, and then now robots. I mean, it seems like this, this is a very exciting world of research that I would love to see more people get into. So congrats on on this work. I'm I'm really excited to follow along. Thanks. Yeah, Speaker 1 24:12 I think that the it's the ability for more people to investigate this kingdom is going to be why it proliferates faster and has more exciting results. I believe it's just very difficult. It costs millions of dollars and years and years of time to work with bio mammalian tissue systems, and the benefit is actuation, but if you try to do anything else, I think that the fungal kingdom, as well as plant kingdom, offers a lot of benefits to work with. We like, we there. So Barbara Matsui is a, is a, actually, we're. And former PhD advisor in Italy. She works on really great plant interfaces, but we like fungus because they culture really fast when you can get it to work. So the results come faster. You can template their growth really well. So the future for us isn't just going to be a Petri dish in the middle of the robot. The robot itself is going to be confluent with the mycelia, and we think that'll allow the persistent and localized chemical monitoring for the for the response. We're probably going to not just work on mobile robots over the ground, but actually mobile robots underground that have the these sensing capabilities as well, but we are very focused on agriculture as our as our target for utility. Lera 25:52 Right, right in, in, in the paper, there was a sentence that I would, I would love to learn more about, is that third scenario, the researchers were able to override the mycelious native signal entirely. What does that mean? What does it mean to to override the mycelias native signal entirely? Speaker 1 26:12 We so this is a three Anand and I had different perspectives on how to what to do with this paper. Anand was and I think it was a good idea in the end, at the end, to use the just the action potentials measured from the mycelia to directly control the motors and the valves on the the two different types of robots, the walking one used valves, the wheeled one used rotary motors, but And so the reason that's a good idea is because that is more towards the signal processing that'll be required to measure the environment and then provide stimulus to it. For a mobile robot, it just kind of does something, and what I wanted it to do is to move faster in a more logical way. The logic was light on move at a speed proportional to the amplitude and duration. And so then what overriding the signal meant and and correct me if I'm wrong, because I think actually you wrote this. Actually you wrote this, this, this portion of it, it's hard to say we there's so many edits back and forth about about it, but that to what that means to me is that instead of directly correlating the action potential sequence to the motor control, there was a period around 30 seconds of the weight potential waveform that was then correlated to a function that the robot then did, and then there's a sliding window according to that. So even though that's why, when you see a light on it, it doesn't immediately do something. It's measuring a period and then making a decision of what to do based on that. Have you seen Lera 28:06 I watched this video years ago, so I can't remember the exact information, and maybe, maybe you know what I'm talking about, but, but some researchers were using AI, and I don't know the exact term, data learning maybe is the right term, but the robot was able to learn new tasks that the researchers didn't program. And so they figured out that the robot was using facial recognition. And the researchers were like, We never programmed it to to use facial recognition, but it just was using machine learning to to come up with new, new ways to evolve. And I thought that was pretty crazy. I don't know if you're using any sort of machine learning or AI to to kind of help this process. Speaker 1 28:58 I think, I think you were searching for deep learning. Maybe deep learning. Yeah, different layers. I'll let Ani was much more involved in the signal processing than I was, so maybe I'll just, I'll fill in stuff if there's anything legal on it. Yes. Speaker 2 29:15 So, so I think yeah. So in this case, we did not use in machine learning. We use very like standard statistics, you know, where you can have, like, you have these coordinate spiking that is coming from fungi, and you want to make sure that you can detect it, because these spikes are have like, you know, sort of like a height and width information. So we did not, we just use a standard statistics to, you know, measure that, and then we just try to convert it to some sort of a digital business. I mean, maybe in the future years, if we are interested with doing that. But definitely not, not in this paper. Lera 29:55 So if you guys, I know you said you want to get into agricultural application. Applications. But if you had unlimited money, Team resources, equipment, all the permits you ever needed, every species of fungi at your fingertips, what would you do and why? Speaker 2 30:17 So if you ask me, I want to send this to another planet I wanna, I wanna leverage, like, these robots. Like, you know, there's a kind of, if, you know, there's a term called muscle and sleep control in robotics, where you have a, like, a central control and then you have a lot of, like, you know, distributed control system. So something like, you know, I envision that, oh, we can have a robot which is locally controlled by my Celia. But if something want to do, we can impose it from outside, so we can send it to, probably, I don't know, Europe or something, to do such some light or something. That's what, that's what I would do. Of course, I will first find it out if this, I mean, of course, there are ethical other concerns, like if you can send anything like that, but if you have the flexibility and opportunity you want to do that? Yes, I Lera 31:07 mean, there's a theory called panspermia, that fungi came from space and colonized the planet from an asteroid. So maybe, maybe we're just sending them back home. Who knows? I How about you, Rob, Speaker 1 31:26 I like the idea of having the organisms compete to emerge at a solution that wouldn't happen naturally like you can if you were to put well, let's just say different species of fungus and your The goal was maybe you template their growth using an electric field bias as well as a chemical gradient. You might be able to get nano architectures that are useful for catalysis or high power density chemical power sources. Basically, animals are really power. They're agile and they're enduring. They can do high power motions, and they can do it for a long time, way better than any of our machines can. And it's because they use Chem, energy, dense chemicals, to fuel themselves. The problem with that is you need mass transport. You need to get oxygen and sugar throughout the body to all these different things. And how nature does it? Again, it's this bottom up versus top down thing. They grow capillaries and have a very high surface area to volume ratio that allows them to put these chemicals everywhere. Fungi do do that too. They they create these very the fractal, nanoscale networks. And I think we can template that growth to allow the same thing so they could be the capillaries that like like they do. I mean, in nature, mediating plant growth, or trees and these kinds of things. They could maybe do the same thing for the different modules in robots, the control actuator and power subsystems. They could be the regulatory body that controls all that, and will allow us to have this solve this compromise between endurance and agility. So that's what and at the same time, every, every hyphy is a signal processing unit. It's measuring it's self powered. There's it's got its own power. It's measuring something, and it's sending a signal. And if it's like having a million sensors that you don't have to send wires to. So anyway, yeah, I think sense of sensorized high surface area to volume ratio structures is, was what we would do, Lera 34:20 yeah, and it makes sense more on ins, yeah, they're, they've been on this planet for a billion years plus, and it had to adapt pretty rapidly for all different ecosystems. And so they're supreme pharmaceutical factories that we, if we can tap into and use biomimicry to solve some of our world's biggest problems, then you know their best ally to do it cool. I don't want to make you late to your meeting. Thanks for pushing that back. Where can people follow your work and keep updated on the robotics processes? Speaker 1 34:56 Thanks so much, Alex. This is like really fun, and you're so. Informed and smart. So I really, Lera 35:02 really appreciate. I was like, Oh man, I Speaker 1 35:05 gotta fight to have more time here. But it was, yeah, my our group website is O, R, L, dot, M, A, e.cornell.edu, we're actually spinning back up our Instagram account, which has been dormant for a while, and then, so I'll send you that too, but yeah, just the website for now. And thanks a lot, sweet. Well, thank Lera 35:28 you guys, and thank you for everyone tuning in and trimming in for another episode of the mushroom revival podcast. If you like the show you want to support, we have a mother brand, mushroom revival. Mushroom revival.com we have a whole line of organic mushroom supplements from capsules, tinctures, gummies, powders. We have a coupon code just for listeners, and that coupon code is pod treat. We also have a bunch of free things on our website, including a giveaway. The link is in the in the bio. We pick a winner once a month. We also have a whole hundreds of blog posts. We have over a dozen ebooks that you can download for free. And my newest book, The Little Book of mushrooms, is on there as well. And keep spreading the word. If you've learned something cool in this episode, tell a friend. Tell a random person checking you out at the grocery store. Get people excited about fungi and nature and new technologies coming out, and give people hope in this world. So with that much love and may the spores be with you. Applause, Transcribed by https://otter.ai
