**Biomimicry in Robotics: A Visionary Future**

As we explore the realm of robotics and biomimicry, it becomes increasingly clear that nature has much to teach us about innovation and design. The latest breakthroughs in this field are yielding robots that mimic the movements and behaviors of animals, with the ultimate goal of creating machines that can adapt and interact with their environment like living organisms.

**The Vision for Room BOTS**

One of the most exciting developments in biomimicry robotics is the concept of Room BOTS. This vision envisions a future where furniture is not just functional, but also interactive and adaptable to our needs. Imagine waking up in the morning to find your bed has been transformed into a gel-like structure that can change shape to accommodate different guests or activities. This is the kind of futuristic thinking that Room BOTS promises to deliver.

**Meet Simon and Joshua from EPFL**

We had the pleasure of chatting with Simon, a member of the junior team at EPFL, who shared with us their experiences working on biomimicry robotics projects. "We're learning from real-world animals movements," he explained, "and trying to apply that knowledge to create robots that can move like snakes or quadrupeds." When asked about their favorite robot design, Simon mentioned the lamprey robot, which impressed him with its simplicity and lifelike motion.

**Cheetah Cub Inspiration**

Another standout example of biomimicry robotics is a cheetah cub-inspired robot. This little marvel uses a pulley system to achieve remarkable speed and agility, much like its real-life namesake. The researchers who created this robot explained that they were inspired by the natural world and sought to capture the essence of animal movement in their design.

**Self-Assembling Robots**

In another exciting development, self-assembling robots are being explored as a potential future technology. These modular robots can be combined in various ways to create complex structures or even furniture, much like the Room BOTS concept. The idea is that swarms of these small robots could come together to build and assemble things on their own, using algorithms and simulations to guide them.

**Swarm Robotics: A Future Possibility**

As we look ahead to the future, it's clear that swarm robotics has the potential to revolutionize the way we design and interact with machines. By harnessing the power of swarms, these robots could potentially be used for a wide range of applications, from construction to healthcare. The possibilities are endless, and it will be exciting to see how this technology develops in the years to come.

**Conclusion**

Biomimicry robotics is an exciting field that holds great promise for innovation and design. By studying the natural world and mimicking animal movements, researchers can create machines that are more efficient, adaptive, and responsive. As we look ahead to the future, it's clear that robots like Room BOTS, cheetah cub-inspired robots, and self-assembling robots will play a key role in shaping our world. We're excited to see where this technology takes us and what wonders it will create in the years to come.

"WEBVTTKind: captionsLanguage: enhey everybody its norm from tested and I'm so excited to introduce you guys to this guy right here is Kishore Hari he's a director of the Bay Area Science Festival now I'm not interviewing Kishore because we're going to be interviewing some people together we have an amazing line of people we're talking to about robots robots bio robots now we're here at the Swiss next Institute they bring organizations and researchers from Switzerland and tonight we're speaking the people from the EPFL EPFL one of the major science institutes in switzerland and we have a special biomimicry robotics group that came over that model all sorts of behavior of snakes and fish and salamanders and reptiles all to understand how biology from nature can really inspire the next generation of robotics my robots lovable or creepy let's find out hi I'm here with Alka a spirit and to talk about these wonderful biomimicry robots you have who do we have on display here so this is some fee but it's an eel like or lamprey like robot so remind us what is a lamprey versus an eel yeah so lampreys are very primitive vertebrates it's one of the oldest retro baits existing so it's just still a fish eel it's a bit more advanced it both are fishes but they both show exactly the same type of swimming it's a basic sort of undulation through the water absolutely it's cool I'm really firm you have a nice traveling wave undulation propagated from head to tail and why did you want to study a lamprey motion you said it's one of the most ancient animals what can we learn from something like that yet up to main purposes the first one is for neuroscience it's like a tool for neuroscience to understand the spinal cord of the real lamprey because it's probably the animal of which we know best the spinal cord so we use it as a testbed basically to test hypotheses of how it's organized meaning that a lot of neuro scientists have been studying that lamprey for a long time especially the spinal column absolutely like does it have like a brain like how we associate brains and other animals or is a lot of this controlled through the Chinatown well that's the interesting part it has like any vector beta part of the brain and it has a spinal cord but for locomotion is really the spinal cord which is a key element and people know it like you can cut the head of a chicken can still run for a while because the spinal cord is really the controller for locomotion and even though it's simple motion how complex of a of a creation is this robot well the big challenge was to make it waterproof there we suffered a bit so we had some generation to make it really waterproof otherwise what's nice it's a very modular design so we we have little elements every module has its own micro controller battery and it has a kundus to connect them so we can very easily make the robot longer shorter or even add legs later on we will see a salamander version of this robot and how long our lamprey is usually so lamprey an adult is like one meter one more yard yeah and when you are testing this robot are you looking at that specific 1 meter length are you looking at even longer lengths or even shorter links to module to model this this kind of yes s wave pattern for us because you as you'll see the we have some kind of granularity problem like the fact that the modules cannot be too small it's good to be at least 1 meter otherwise you cannot really propagate properly the traveling wave and how long can this thing go in the water how long it can swim yeah it probably for half an hour non-stop easily proxy if you have a few stops for one or two hours and so tell us about the intelligence in this like I it's one thing for this to swim through we've all seen those toys that swim through the water they're just going through the same motion what are we seeing inside here in terms of the intelligence yeah the intelligence is is really just kind of replica of the spinal cord circuits and what's nice it's a very distributed system so you can almost cut it in half and subpar to the smile cord can still work and not animal the way the animal would die it's not like an invertebrate like a leech or but the lamprey with diabetes can really it's very distributed control and let's say and what's interesting it's very robust so you can really modulate it per cube per disturb it a lot and what might make the most fascinating is the signals coming from the upper part of the brain being sent to the spinal cord are very simple just two signals you increase both of them you go faster use you stimulate more one sign and you will turn towards outside so you see a demo when it swims is very easy it's as if you had a remote-control car like just two singles to control the whole locomotion and I think that's nice because that means the upper part of the brain doesn't need to worry about muscles what every muscle should do it's really the job the spinal cord to do that and an upper part of the brain will just do a high-level modulation and so this controller basically has what you said just as two degrees of motion absolutely so Alessandra Krispies here the postdoc in the lab who who developed the electronics of the robot it's basically playing the part of the upper part of a brain let's say and then he sends this very simple signals to the spinal cord model and that's running in real-time now with little mathematical models onboard of the robot how complicated a mathematical model is it to mimic this behavior it's a fairly simple like the complexity comes from coupling the oscillators together so it's a network of coupled oscillators from from a mathematical point of view what is the next generation of this look like so we have a new version which is longer a more rugged for for unlike for field like a field robot so we use it for pollution detection so we have a GPS on board we can do navigation so be autonomous I can absolute we can unleash a number of these throughout a water system yeah you can give waypoints it will go to the waypoints or more interestingly it should be able to find a gradient like climbing where the pollution comes from an even at sensing equipment I mean we've seen GPS get that small but the actual chemical detection can be miniaturized into something this small yeah that we have colleagues in Lausanne that work exactly on that so that's not our job it's their job but they do a very good job for some of these sensors to to have them on board absolutely oh this is a cute little guy yeah yeah yeah I have a lot of fondness for him you do you have a fondness for new yeah absolutely it you know you know what this one of my still my favorite robot that we have in the lab and one of the reasons is that if you take it out the appearance looks like nothing it's just the set of boxes connected together but what makes it cool is the motion it's the motion that makes it lifelike so easily we could make a skin over it too to improve the appearance but I like this notion that the appearance is not so beautiful but the motion is beautiful thank you for showing us the lamprey you're most welcome and so from water-based systems we move to land-based systems where locomotion is a little bit different but snakes we have less so this is camilo and tell us about your snakes okay well this is negative robot that are very simple let's say in the way they're constructing for something we have one motor that have only one degree of freedom then we just twist it motor to have another degree of freedom like this and then we just continue with the same chain doing all over the number of models that you want to put in terms of biomimicry you're trying to mimic the Versa tone in emotion yeah mistake a snake doesn't actually have just you know one degree in turn or does it not doesn't I mean depend of what game do you try to reproduce for example you can see that actually you the more the body moves like a real snake by us and we have also this plant this vertical plane so you can move it in very dimensions so depending on the Gator try to do for example syusai wine and snakes you can just represent the real scientists when you study a land snake right yeah and it's gates you know it can slither forward like you say can move sideways it can coil which are the motions that you find most interesting for this type of robot yeah there is one more shoe that is quite interesting is this one you can see this is called rolling so basically what we are doing is just try to maintain this arch shape all the time but the robots are moving all the time like this so every model just move like you combine this motion you have a rolling so that's very interesting because for example you can wrap around something like a pony like a tree like a tree branch for example and then you just roll and you can just progress so tell it's traversing along other long objects basically some type of a poor girl in the ground or on the ground and have a little bump you kind of steep pass over right and so when you're designing something Chris it's not just illustrate how you can use engineering mechanics to mimic the biology of an animal but there's also practical application yeah it's not just making robots takes for robots takes that's high around foreshores neck that is a well with a practical occasion that we most information is for example going to a search-and-rescue mission right so you know we have this cross sectional so basically this is super slim so you can put it through a penny board hole that you can imagine like this size and then you can just once inside you can just move and do a lot of things right so as I mentioned we can have a different types of gait so it doesn't matter how the terrain is how the looks you can strike I still try to go in this gate and then just change to another one so this is what we call multimodal locomotion this is pretty let's say convenient for this kind of a scenario and imagine playing a camera on a robot like this very useful for search and rescue now this is scale I mean you add more links to it make it a longer stick that changed how they have scored for example I'm going to show you something we have here a degree of freedom that's in done in eight modules it's quite a small name but still we can do a lot of things this big one over here it has two this one is 60 and 2424 so you can see that kind of scale the only and let's say drawback is that as you scale the size you need more battery power so in that case you need to put for example in the hip we have this controller one here running and in the other one we have the battery so you scale it even more you need a little a large room let's say for the battery but the software also scales a properly exactly but it doesn't get more complicated doesn't know that all ties socialite simple because basically you just control two waves while vertical and one horizontal wave and that's it that's amazing yeah well the snakes don't have legs but we're take a little look at robots that's do but thank you so much Shane would let's come the snake robots Whitehall it absolutely whoo from snakes to mammals oh yeah what do we have here so this is little cheetah cub it's a bio-inspired robot modeled after the the animal a cat or a dog so quadrupedal memo tara so memo the cool thing about this robot are actually its legs so you have here these free segments one two three and this string kind of put the string connected exactly so one motor that's sitted here that's sitting here that's doing the front back movement and one motor that is compressing the leg so for that I have to hold it here a little bit like compressing here the lac and the spring is extending how much is that like how our arm moves actually quite similar because you have this antagonistic motion so you have always one model that muscle is contracting in one muscle that is protracting and exactly that's what the motor does Motors contracting and the spring is retracting again this seems like a very robust system how fast can a creature like this move Oh with what is like you know seemingly simple to two motion system yeah that is actually the Forte of this of this robot it goes up to seven body lengths per second seven body links per second so let's translate that into into some into symmetric units it's a metric in that video is good so that's 1.4 meters per second oh that's pretty fast yeah that's a button out round that by much that's about like like a human would fast walk yeah so that's a maximum that it can go that's pretty that's exceptional just with this this simple gear system in expect especially because it has no sensors on it so this one is completely open loop doesn't know about the environment and these intelligent mechanics they are actually stabilizing the motion so what happens when we start to add sensors to it what does it start to look like then it starts to look like much like this one here so this is on chilla they have a lot more it's a lot bigger a lot heavier I see that it sucks yeah and for here we added two acts for aesthetics no but that's actually for protecting three D four sensors that we have on the feet so this is freely contact sensors in the feet we have encoders on each joint of the of the whole thing the legs are exactly the same principle you have again the protraction redirection and the compression decompression when you say there's four sensors in the feet what is it really trying to to sense like how how stable the ground is below it how what the shape exactly so you have a little bit the the possibility to go into reflexes so for example Istanbul stumbling reflex when the when the foot hits obstacle and the robot would normally stumble it retracts the the leg even more and like this moves over the obstacle and that's something that we can sense with these sensors what kind of processor do you have to have inside of this - in real time monitor all these sensors actually not much there's a one kid or a one yeah one gigahertz sensor on the one gigahertz processor on it we're using CPG networks they are very low compared that they have a very low computational cost so and to add that sense to that sensory feedback doesn't cost much in computation as long as it is feedback not like this another camera or something camera imaging costs a lot of computational can we see it actually move yeah we can so let it run in the air a little bit and then so the adduction abduction as you can okay so it's starting to run a little bit in your hand in my hand what does that feel like on your hand how much forces ah it's a hard to control you can't you can't feel it I mean that there's this you can feel kind of leadership yeah but Galax actually very low in their time imbalance in the in general let's balance the robot is mostly symmetric so left-right for back it's quite symmetric so it has a good stability due to this construction already so that's just ask a girl and it's about 0.7 meters per second so it's seller to the to the previous one yeah half the maximum speed of the previous one so is that a limitation of the actual technology using this sort of like flee system or is it the limitation of the battery or something else that's a it's a little bit limitation of the of the power that you can put into the motors because we drive it all on a battery but also on the same time it's a it's a very safe economic way to drive the robot so at this speed we get the most runtime out of the battery let's say so why we let it on the floor and see it as you move around you go oh I had it backwards up ahead and on the cable so besides the the 3d touch sensors in the feet what other sensors are on board so there are encoders in each joint backward encode us in each joins yes an I'm you on it which you can have the upholster control with it is it stabilized or will it fall over it's quite stable from the from the incentive mechanics already so this springy lag it's putting the robot in a very stable position general what are the applications of building a cheetah like this what are you really trying to model in terms of uh locomotion so in the look in the locomotion it's it's very interesting because uh you can see how good the mechanics are already before the actual control kicks in so you outsource a little bit of the of the control effort in the mechanics itself like this you gain a lot of computation space let's say for higher computation tests like like vision when you would want vision on it or something like that thanks for sharing us the cheetah yeah thanks for talking to us about it we are very excited and finally robots that may not exactly be inspired by a real animal in nature but something that maybe takes lessons from the previous robots I mean tell us about this Roomba this modular robot that exactly it's a modular robot this is one unit it consists of two inseparable spheres that have three degrees of freedom that can continuously rotate like this and in those two diagonal directions then each of those modules has ten attachment planes and in some of them we have hooks retractable claws that come out like this oh wow and this enables one module to connect to other modules and form a bigger structure than module itself but they are self-sufficient just in this pair exactly everything is in here we have battery in here we have all the sensors for our motors in here and we have communication over bluetooth to our centralized PC now locomotion is a theme that we've gone through we saw from the lamp right basic locomotion and water all the way to a quadruped read this one doesn't have legs so I imagine it moves a little more like the snake robot we saw earlier so here's the cool thing about them we can make them move like snakes but we can also configure it into a different robot and make it move like a quadruped it this was it can be everything what you want them to yes so this is we connect them how we want them we play our control and we can make them move like an animal or like a snake or however you would like them to be right how they latch together I imagine is important because it needs to be a strong structurally how many points they can latch together now how did you decide upon one you needing two units to be a singular module and also the axis of movement so this is just see it is a little more more complex module than there from the snakes mm-hmm so we decide and they had one degree of freedom yeah now we have three degrees of freedom that allow them each module by itself to be a little more free than the the others are and I don't know they look cool like this with diagonal and it's an experimentation your figuring out that happens will you add two more degrees of movement when you have the algorithms from the other animals that you can configure to make this wiggle to make it crawl to have them attached without each other how many of you actually put together so right now we are in the process of building more the furthest we went this maybe four or five modules we made this the structure stronger the gear stronger so now we're hoping to put some more together maybe eight to ten something like this we are in the process of doing exactly right hands between actually building modules it's just lots of simulations right oh yeah definitely yeah we have to make sure them in simulation our mode of modules don't collide with each other if it will happen in the real robot unfortunately they would immediately break so we need to we need to amass a lot of algorithmic simulations for this do your your team imagine that this is something that can be used not only in the field but also indoors and can get hard and miniaturize definitely so the name room bot comes from the vision to have instead of furniture you just have a bunch of those in your room and you you configure all your furniture with those room BOTS so imagine you wake up in the morning in a bed out of room BOTS they config into a gel in tables if you have more guests you just take a bunch of those more and they lookin fit into sofas into your living space so this would be this would be the future vision plane and you're taking lessons from real world animals movements and other animals and snakes and quadrupeds very and it's going to make a beautiful modular chair in the future that's awesome thank you so much Simon you're like a junior team at the EPFL it's been great chatting a few guys and we love your robots thank you Joshua that was awesome adorable turns out incredibly adorable I think I want to adopt one yeah the lamprey that's that's what I put in a bathtub yeah I love the lamprey and I loved how simple that motion was but how lifelike the what it wouldn't look like and felt like it was almost like I could I fell in love with it it was it had emotions to it as it was swimming through the water but what really caught my eye was the cheetah cub it was just that simple motion that sort of back-and-forth motion how they could actually get you know basically like normal walking speed out of a simple what looked like a pulley system I remember those like transformers toys I used to have and they would have just the motors up here and kind of stutter around it had it that same sound to write those robots but what's way more elegant way weren't looking and beautiful how the gate sort of slowly moved through what about the self-assembling robots the self-assembly robot it's kind of like the thing that's far out in the future Futurama Jetsons thing they want to combine these modular robots and they're big right now but they can make it smaller and haven't you can group dozens of them together eventually swarm them and have them build chairs and furniture well I think the swarm of the self assembling robot sounds a lot better to me than the swarm of snakes right yeah but you know what they're all going to move like snakes you gotta go from two to a snake than a chair and then it's okay the progression of locomotion was incredible so I can't wait to see more robotics in this biomimicry realm that allows us to use that term nature inspire to do everything from climbing walls to maybe even flying and we got to thank Swiss necks for inviting us here thank the EPFL and their researchers and thank you cash or for bringing us here as well and chatting with some of the researchers we'll have more cash or on testing we're gonna go to really interesting science Institute's meet more robots talk to other researchers and scientists a lot more science and so stay tuned for that on tested calm until then we'll see you guys next time see ya righthey everybody its norm from tested and I'm so excited to introduce you guys to this guy right here is Kishore Hari he's a director of the Bay Area Science Festival now I'm not interviewing Kishore because we're going to be interviewing some people together we have an amazing line of people we're talking to about robots robots bio robots now we're here at the Swiss next Institute they bring organizations and researchers from Switzerland and tonight we're speaking the people from the EPFL EPFL one of the major science institutes in switzerland and we have a special biomimicry robotics group that came over that model all sorts of behavior of snakes and fish and salamanders and reptiles all to understand how biology from nature can really inspire the next generation of robotics my robots lovable or creepy let's find out hi I'm here with Alka a spirit and to talk about these wonderful biomimicry robots you have who do we have on display here so this is some fee but it's an eel like or lamprey like robot so remind us what is a lamprey versus an eel yeah so lampreys are very primitive vertebrates it's one of the oldest retro baits existing so it's just still a fish eel it's a bit more advanced it both are fishes but they both show exactly the same type of swimming it's a basic sort of undulation through the water absolutely it's cool I'm really firm you have a nice traveling wave undulation propagated from head to tail and why did you want to study a lamprey motion you said it's one of the most ancient animals what can we learn from something like that yet up to main purposes the first one is for neuroscience it's like a tool for neuroscience to understand the spinal cord of the real lamprey because it's probably the animal of which we know best the spinal cord so we use it as a testbed basically to test hypotheses of how it's organized meaning that a lot of neuro scientists have been studying that lamprey for a long time especially the spinal column absolutely like does it have like a brain like how we associate brains and other animals or is a lot of this controlled through the Chinatown well that's the interesting part it has like any vector beta part of the brain and it has a spinal cord but for locomotion is really the spinal cord which is a key element and people know it like you can cut the head of a chicken can still run for a while because the spinal cord is really the controller for locomotion and even though it's simple motion how complex of a of a creation is this robot well the big challenge was to make it waterproof there we suffered a bit so we had some generation to make it really waterproof otherwise what's nice it's a very modular design so we we have little elements every module has its own micro controller battery and it has a kundus to connect them so we can very easily make the robot longer shorter or even add legs later on we will see a salamander version of this robot and how long our lamprey is usually so lamprey an adult is like one meter one more yard yeah and when you are testing this robot are you looking at that specific 1 meter length are you looking at even longer lengths or even shorter links to module to model this this kind of yes s wave pattern for us because you as you'll see the we have some kind of granularity problem like the fact that the modules cannot be too small it's good to be at least 1 meter otherwise you cannot really propagate properly the traveling wave and how long can this thing go in the water how long it can swim yeah it probably for half an hour non-stop easily proxy if you have a few stops for one or two hours and so tell us about the intelligence in this like I it's one thing for this to swim through we've all seen those toys that swim through the water they're just going through the same motion what are we seeing inside here in terms of the intelligence yeah the intelligence is is really just kind of replica of the spinal cord circuits and what's nice it's a very distributed system so you can almost cut it in half and subpar to the smile cord can still work and not animal the way the animal would die it's not like an invertebrate like a leech or but the lamprey with diabetes can really it's very distributed control and let's say and what's interesting it's very robust so you can really modulate it per cube per disturb it a lot and what might make the most fascinating is the signals coming from the upper part of the brain being sent to the spinal cord are very simple just two signals you increase both of them you go faster use you stimulate more one sign and you will turn towards outside so you see a demo when it swims is very easy it's as if you had a remote-control car like just two singles to control the whole locomotion and I think that's nice because that means the upper part of the brain doesn't need to worry about muscles what every muscle should do it's really the job the spinal cord to do that and an upper part of the brain will just do a high-level modulation and so this controller basically has what you said just as two degrees of motion absolutely so Alessandra Krispies here the postdoc in the lab who who developed the electronics of the robot it's basically playing the part of the upper part of a brain let's say and then he sends this very simple signals to the spinal cord model and that's running in real-time now with little mathematical models onboard of the robot how complicated a mathematical model is it to mimic this behavior it's a fairly simple like the complexity comes from coupling the oscillators together so it's a network of coupled oscillators from from a mathematical point of view what is the next generation of this look like so we have a new version which is longer a more rugged for for unlike for field like a field robot so we use it for pollution detection so we have a GPS on board we can do navigation so be autonomous I can absolute we can unleash a number of these throughout a water system yeah you can give waypoints it will go to the waypoints or more interestingly it should be able to find a gradient like climbing where the pollution comes from an even at sensing equipment I mean we've seen GPS get that small but the actual chemical detection can be miniaturized into something this small yeah that we have colleagues in Lausanne that work exactly on that so that's not our job it's their job but they do a very good job for some of these sensors to to have them on board absolutely oh this is a cute little guy yeah yeah yeah I have a lot of fondness for him you do you have a fondness for new yeah absolutely it you know you know what this one of my still my favorite robot that we have in the lab and one of the reasons is that if you take it out the appearance looks like nothing it's just the set of boxes connected together but what makes it cool is the motion it's the motion that makes it lifelike so easily we could make a skin over it too to improve the appearance but I like this notion that the appearance is not so beautiful but the motion is beautiful thank you for showing us the lamprey you're most welcome and so from water-based systems we move to land-based systems where locomotion is a little bit different but snakes we have less so this is camilo and tell us about your snakes okay well this is negative robot that are very simple let's say in the way they're constructing for something we have one motor that have only one degree of freedom then we just twist it motor to have another degree of freedom like this and then we just continue with the same chain doing all over the number of models that you want to put in terms of biomimicry you're trying to mimic the Versa tone in emotion yeah mistake a snake doesn't actually have just you know one degree in turn or does it not doesn't I mean depend of what game do you try to reproduce for example you can see that actually you the more the body moves like a real snake by us and we have also this plant this vertical plane so you can move it in very dimensions so depending on the Gator try to do for example syusai wine and snakes you can just represent the real scientists when you study a land snake right yeah and it's gates you know it can slither forward like you say can move sideways it can coil which are the motions that you find most interesting for this type of robot yeah there is one more shoe that is quite interesting is this one you can see this is called rolling so basically what we are doing is just try to maintain this arch shape all the time but the robots are moving all the time like this so every model just move like you combine this motion you have a rolling so that's very interesting because for example you can wrap around something like a pony like a tree like a tree branch for example and then you just roll and you can just progress so tell it's traversing along other long objects basically some type of a poor girl in the ground or on the ground and have a little bump you kind of steep pass over right and so when you're designing something Chris it's not just illustrate how you can use engineering mechanics to mimic the biology of an animal but there's also practical application yeah it's not just making robots takes for robots takes that's high around foreshores neck that is a well with a practical occasion that we most information is for example going to a search-and-rescue mission right so you know we have this cross sectional so basically this is super slim so you can put it through a penny board hole that you can imagine like this size and then you can just once inside you can just move and do a lot of things right so as I mentioned we can have a different types of gait so it doesn't matter how the terrain is how the looks you can strike I still try to go in this gate and then just change to another one so this is what we call multimodal locomotion this is pretty let's say convenient for this kind of a scenario and imagine playing a camera on a robot like this very useful for search and rescue now this is scale I mean you add more links to it make it a longer stick that changed how they have scored for example I'm going to show you something we have here a degree of freedom that's in done in eight modules it's quite a small name but still we can do a lot of things this big one over here it has two this one is 60 and 2424 so you can see that kind of scale the only and let's say drawback is that as you scale the size you need more battery power so in that case you need to put for example in the hip we have this controller one here running and in the other one we have the battery so you scale it even more you need a little a large room let's say for the battery but the software also scales a properly exactly but it doesn't get more complicated doesn't know that all ties socialite simple because basically you just control two waves while vertical and one horizontal wave and that's it that's amazing yeah well the snakes don't have legs but we're take a little look at robots that's do but thank you so much Shane would let's come the snake robots Whitehall it absolutely whoo from snakes to mammals oh yeah what do we have here so this is little cheetah cub it's a bio-inspired robot modeled after the the animal a cat or a dog so quadrupedal memo tara so memo the cool thing about this robot are actually its legs so you have here these free segments one two three and this string kind of put the string connected exactly so one motor that's sitted here that's sitting here that's doing the front back movement and one motor that is compressing the leg so for that I have to hold it here a little bit like compressing here the lac and the spring is extending how much is that like how our arm moves actually quite similar because you have this antagonistic motion so you have always one model that muscle is contracting in one muscle that is protracting and exactly that's what the motor does Motors contracting and the spring is retracting again this seems like a very robust system how fast can a creature like this move Oh with what is like you know seemingly simple to two motion system yeah that is actually the Forte of this of this robot it goes up to seven body lengths per second seven body links per second so let's translate that into into some into symmetric units it's a metric in that video is good so that's 1.4 meters per second oh that's pretty fast yeah that's a button out round that by much that's about like like a human would fast walk yeah so that's a maximum that it can go that's pretty that's exceptional just with this this simple gear system in expect especially because it has no sensors on it so this one is completely open loop doesn't know about the environment and these intelligent mechanics they are actually stabilizing the motion so what happens when we start to add sensors to it what does it start to look like then it starts to look like much like this one here so this is on chilla they have a lot more it's a lot bigger a lot heavier I see that it sucks yeah and for here we added two acts for aesthetics no but that's actually for protecting three D four sensors that we have on the feet so this is freely contact sensors in the feet we have encoders on each joint of the of the whole thing the legs are exactly the same principle you have again the protraction redirection and the compression decompression when you say there's four sensors in the feet what is it really trying to to sense like how how stable the ground is below it how what the shape exactly so you have a little bit the the possibility to go into reflexes so for example Istanbul stumbling reflex when the when the foot hits obstacle and the robot would normally stumble it retracts the the leg even more and like this moves over the obstacle and that's something that we can sense with these sensors what kind of processor do you have to have inside of this - in real time monitor all these sensors actually not much there's a one kid or a one yeah one gigahertz sensor on the one gigahertz processor on it we're using CPG networks they are very low compared that they have a very low computational cost so and to add that sense to that sensory feedback doesn't cost much in computation as long as it is feedback not like this another camera or something camera imaging costs a lot of computational can we see it actually move yeah we can so let it run in the air a little bit and then so the adduction abduction as you can okay so it's starting to run a little bit in your hand in my hand what does that feel like on your hand how much forces ah it's a hard to control you can't you can't feel it I mean that there's this you can feel kind of leadership yeah but Galax actually very low in their time imbalance in the in general let's balance the robot is mostly symmetric so left-right for back it's quite symmetric so it has a good stability due to this construction already so that's just ask a girl and it's about 0.7 meters per second so it's seller to the to the previous one yeah half the maximum speed of the previous one so is that a limitation of the actual technology using this sort of like flee system or is it the limitation of the battery or something else that's a it's a little bit limitation of the of the power that you can put into the motors because we drive it all on a battery but also on the same time it's a it's a very safe economic way to drive the robot so at this speed we get the most runtime out of the battery let's say so why we let it on the floor and see it as you move around you go oh I had it backwards up ahead and on the cable so besides the the 3d touch sensors in the feet what other sensors are on board so there are encoders in each joint backward encode us in each joins yes an I'm you on it which you can have the upholster control with it is it stabilized or will it fall over it's quite stable from the from the incentive mechanics already so this springy lag it's putting the robot in a very stable position general what are the applications of building a cheetah like this what are you really trying to model in terms of uh locomotion so in the look in the locomotion it's it's very interesting because uh you can see how good the mechanics are already before the actual control kicks in so you outsource a little bit of the of the control effort in the mechanics itself like this you gain a lot of computation space let's say for higher computation tests like like vision when you would want vision on it or something like that thanks for sharing us the cheetah yeah thanks for talking to us about it we are very excited and finally robots that may not exactly be inspired by a real animal in nature but something that maybe takes lessons from the previous robots I mean tell us about this Roomba this modular robot that exactly it's a modular robot this is one unit it consists of two inseparable spheres that have three degrees of freedom that can continuously rotate like this and in those two diagonal directions then each of those modules has ten attachment planes and in some of them we have hooks retractable claws that come out like this oh wow and this enables one module to connect to other modules and form a bigger structure than module itself but they are self-sufficient just in this pair exactly everything is in here we have battery in here we have all the sensors for our motors in here and we have communication over bluetooth to our centralized PC now locomotion is a theme that we've gone through we saw from the lamp right basic locomotion and water all the way to a quadruped read this one doesn't have legs so I imagine it moves a little more like the snake robot we saw earlier so here's the cool thing about them we can make them move like snakes but we can also configure it into a different robot and make it move like a quadruped it this was it can be everything what you want them to yes so this is we connect them how we want them we play our control and we can make them move like an animal or like a snake or however you would like them to be right how they latch together I imagine is important because it needs to be a strong structurally how many points they can latch together now how did you decide upon one you needing two units to be a singular module and also the axis of movement so this is just see it is a little more more complex module than there from the snakes mm-hmm so we decide and they had one degree of freedom yeah now we have three degrees of freedom that allow them each module by itself to be a little more free than the the others are and I don't know they look cool like this with diagonal and it's an experimentation your figuring out that happens will you add two more degrees of movement when you have the algorithms from the other animals that you can configure to make this wiggle to make it crawl to have them attached without each other how many of you actually put together so right now we are in the process of building more the furthest we went this maybe four or five modules we made this the structure stronger the gear stronger so now we're hoping to put some more together maybe eight to ten something like this we are in the process of doing exactly right hands between actually building modules it's just lots of simulations right oh yeah definitely yeah we have to make sure them in simulation our mode of modules don't collide with each other if it will happen in the real robot unfortunately they would immediately break so we need to we need to amass a lot of algorithmic simulations for this do your your team imagine that this is something that can be used not only in the field but also indoors and can get hard and miniaturize definitely so the name room bot comes from the vision to have instead of furniture you just have a bunch of those in your room and you you configure all your furniture with those room BOTS so imagine you wake up in the morning in a bed out of room BOTS they config into a gel in tables if you have more guests you just take a bunch of those more and they lookin fit into sofas into your living space so this would be this would be the future vision plane and you're taking lessons from real world animals movements and other animals and snakes and quadrupeds very and it's going to make a beautiful modular chair in the future that's awesome thank you so much Simon you're like a junior team at the EPFL it's been great chatting a few guys and we love your robots thank you Joshua that was awesome adorable turns out incredibly adorable I think I want to adopt one yeah the lamprey that's that's what I put in a bathtub yeah I love the lamprey and I loved how simple that motion was but how lifelike the what it wouldn't look like and felt like it was almost like I could I fell in love with it it was it had emotions to it as it was swimming through the water but what really caught my eye was the cheetah cub it was just that simple motion that sort of back-and-forth motion how they could actually get you know basically like normal walking speed out of a simple what looked like a pulley system I remember those like transformers toys I used to have and they would have just the motors up here and kind of stutter around it had it that same sound to write those robots but what's way more elegant way weren't looking and beautiful how the gate sort of slowly moved through what about the self-assembling robots the self-assembly robot it's kind of like the thing that's far out in the future Futurama Jetsons thing they want to combine these modular robots and they're big right now but they can make it smaller and haven't you can group dozens of them together eventually swarm them and have them build chairs and furniture well I think the swarm of the self assembling robot sounds a lot better to me than the swarm of snakes right yeah but you know what they're all going to move like snakes you gotta go from two to a snake than a chair and then it's okay the progression of locomotion was incredible so I can't wait to see more robotics in this biomimicry realm that allows us to use that term nature inspire to do everything from climbing walls to maybe even flying and we got to thank Swiss necks for inviting us here thank the EPFL and their researchers and thank you cash or for bringing us here as well and chatting with some of the researchers we'll have more cash or on testing we're gonna go to really interesting science Institute's meet more robots talk to other researchers and scientists a lot more science and so stay tuned for that on tested calm until then we'll see you guys next time see ya right\n"