 Kaj je? Mao. Ok. Zvukam, da je ljudo. Zvukam, da bo v drugi budi. Zdaj, da sem vzelo. Vzelo, da je ljudo, da je ljudo. Vzelo, da je ljudo, da je ljudo. Vzelo, da je ljudo. ali vse, da se počutim, da me dočutim na CES. To je prvič zelo, kaj sem kanalite, da je jožočen, z vsega konštora Iron Curtain. Komenistija, 74. Moje boss je nekaj dobro in nekaj do riski. I could come here. As a young person I had a landlady who spoke a little bit of Hungarian, a little bit of German, little bit of Italian. Triest used to be and maybe still is American pot for this part of Europe. Nama VR where I joined my stay and I thanked the organizers that they invited me to this place, and as a reward I would try to entertain you during my talk which is a little bit strange, To je najrejstvene resultate. Pozivamo, da je to vseh, različno, vseh težko, je vseh z vrštih možnega in v težko. Zato je nekaj, nekaj nekaj, nekaj, nekaj nekaj, nekaj, nekaj, nekaj počust, nekaj počust, nekaj počust,em za to, če bo v minulijstvo, ali bo niko, da je počkaj, je bo專j, ali je.š Ko bilo, da počkaj 3 megabajte informacij, zelo, si, da je 25 x 3 megabajte informacij v 1,2. zato tudi prišličite film v svoj del, ne, naredite milijon in zelo je tudi o kolektivnih vsev in možnosti, možnosti je zelo pravda, da se počutite. Zelo da bo se počutiti na prinsipulji, nekaj je nekaj nekaj, zelo zelo zelo zelo v kontekst, in maybe at some point you will be able to apply it to your place. Case also. No, these are just a few examples related to my talk. There are some things swimming here, but they are not fish, but they are fish skin cells. They will put into this one half millimeter square, and they swim together. This is another example for how cells behave. When they are two types, the reds move faster and like to stick to each other more than to a green one, and the greens move slower and like to stick to each other. This is our experiment, and it shows that over 16 hours you can get a structure which is almost like one millimeter, which is a large scale on the level of the cells, and I will touch upon this. This is a very old movie, a demonstration of an experiment about population dynamics of mice, so not reds. In part, I will show you this to you. I hope I will be able to stop this. OK, so they have the mice, they have the thousands of mice. First they color them, but then they got tired, so it was impossible to follow. OK, so look at this person. This is how they used to collect data. So I would like to show you what happens if you collect data differently. And this is our first drone flock. A little bit, I mean they move a little bit disorganized way, but they still are trying to catch someone. So they have a mission, they chase an evening, cold evening in the past myself. This has a little scientific. I think this person, who is a professor in London, believes that what she is doing is related to science, but what she is doing is she simulates with her colleagues, of course, she may not be even right programs, collective motion. But of course it is done in an artistic way, and if you ask her, she sent me this on Saturday, two days ago. She got ready with this. She says that collective motion is a good vehicle for connecting a number of features of nature and human nature. It touches us, such a picture, such a movie, and through arts she can go even deeper to some of the aspects of collective motion. Now this is from YouTube. Okay, so I have three main topics, and they are intended to be biological physics and robotics, just like the way this meeting is imagined. The first one is, I would say, the biological physics, because I would talk about forces between cells. This was just published. So I would have just published, to be published, to be written up and already appeared, or accepted. But you see here is what happens in a zebra, a fish embryo after five hours, after six hours. Do you see this in lots of, lots of collective motion? And in fact something happens around this time, and some new features appear around the yolk. They have a yolk, the egg divides. Now there are about 14,000 cells in the embryo, and of course we are interested in embryo development for two reasons. One is that embryonic development is one of the most impressive miracles of nature. We don't understand. We don't understand many things, you know? But this is something which happens in the womb of our wives also, when we develop. And it is clear, many papers start with claiming that collective motion plays an important role in embryogenesis, but few eventually take data. Ok, so we studied this. Ok, I found this three days ago, and I thought it was an impressive picture of a zebra embryo around maybe 30 hours. Do you know how much it takes? It takes 72 hours after the first division of the egg, and after three days the fish swims away. So a lot of well-defined collective motion, orchestrated collective motion has to take place so that every cell finds its own place. In a bit more technological terms, the main statement of the paper is that these prekordal plate progenitor cells, as they migrate, they affect the neuroectoderm morphogenesis. And this happens, the most important, these are just names of cells. The biologists know these. The view doesn't have to know. The biologists have to know that these are two kinds of cells which move together during, not together, by the way, in the opposite direction, during embryogenesis. And as it is written here, and these cells, which move against the neuroectodermal cells which flow around the egg, they move in the opposite direction and through forces, which are just regular physical forces, friction forces, they're sticky, they stick to each other more to its kind than to the other kind, but they still have a friction among them. And it is mediated with the e-catterin, which is a regular molecule for these processes. So these are the neuroectodermal cells and they are looking now from above. And here it is from the side. Here is the yolk. This is the white type, the regular, and this is a mutant that doesn't have these PPL cells. So this is what happens if you have these two morphogenesis processes. We will see that by the end this remains structured. So this is the first structural footprint of the brain of the zebrafish. And this one loses this structure because the PPL cells which through biophysical force friction do not modify the way it is distributed in the brain, in the developing brain. I don't go through this because of a lack of time, but I just demonstrate that our part in this huge study was only make the flow fields from the movies and calculate all sorts of time and space dependent correlation functions which would support the idea that the development eventually goes along the way that I already mentioned. Until then you only have some movies. If you do all the calculations and you do much more, five more carefully selected very complicated control experiments then it becomes accepted into nature cell biology. So we extracted these flow fields with the PIV method which is very common to use for these purposes and you see how much different the two kinds the one which doesn't have the PPL and the vietype, streams of cells look like. I move to rats. There will be two experiments or rats, two kinds. Marty and I, who is sitting here has been very deeply involved in both. By the end we both, and the whole group, there were more people, just fell in love with the rats. They are amazing, intelligent, sensitive little animals. We put, this is one of the main questions of collective behavior, that if you act collectively, if there is a little society, members, a group, it's a question whether you are individually more efficient by the end as a part of a group, then if you are not part of a group, you are solitary only. And this is not an earth shaking or very extremely counter-intuitive result that we had obtained. But here, OK. Marty built a maze where rats are usually put into a maze individually and there are lots of publications and experiments about how rats find the way out from a maze. Here a labyrinth. This is our labyrinth. This is the entry and because of the size of the rats, now we could have only a 2 by 4 labyrinth which is not extremely complicated because it would have taken a space which is much larger than a room. So you will see, this shows the tracking with the colors as a function of time, the trajectory of a single. So it started out with red here and then it spent a lot of times between the two extreme cases around here when you see the green color. And this is the shape and this is the topology of our labyrinth. It is hierarchical and there is only one place around here where at the dead end there is a water available. At the other places, they have a similar construction but no water comes out from that places. OK, so this is what I mentioned. Apparently they liked Marty even more than they liked me because he has two of them but if you look at this picture, I couldn't tell who is more curious about the other. This is a movie of the experiment in real time. I will stop it soon. This is just for you to demonstrate that they are relatively fast and at the same time slow they are finding the place where the water is here. So they are searching. And if the movie is five times faster then at some point one of them finds the water and amazingly very soon a number of other rats show up at the same place although it is not as simple as you think. It is very far from how ants do this thing. Ants are also extremely efficient in locating sources of food and then go there but they have a very special mechanism for telling this to the other ants. Rats don't do that. They tend to follow another rat a little bit but if the other rat is going back to the water drink a little bit, coming away, going back I cannot present you all of the possible explanations and calculations but the conclusion is this that if you order eight rats in the order in which they got access to water so some found it very quickly like here this is an individual which found the water the second fast his time was just one so you order it along the time how much it takes them to find water if they are in group this is individual the most inefficient rat here found it around 300 seconds after being individually entered into the maze if they were entered into the maze in a group then the last one who got to the resource, the water for that it took only 180 seconds so the group as a whole was at least two times more efficient and we excluded things like smell or visual clues everything was excluded Matej was very careful about this he even had arguments about I would have liked to help the rats to communicate somehow but they didn't communicate in any ordinary way it was just motion this is something which we should write up because we have so many interesting results that we are behind writing up the paper this is a several years long study reminiscent of the one which I showed you from 1962 but instead of binocular we used other methods we had four groups over seven individuals in four arenas everything was recorded over ten months because I will tell you why so 30 terabyte data accumulated in the form of video data we tested things night versus dark daylight mixing group members this was exciting our groups behaved very differently they were all from the same origin we ordered them from a provider they brought the four times seven rats to us we paid the fee and we put them into four groups and the four groups behaved completely differently and then we mixed the groups and after mixing the groups started to behave again very differently and this makes writing up the paper more difficult but it shows that the process is extremely rich and the complexity of social life which we are subject to may have some origins already on the level of rats and most of the experiments on collective behavior are not done on mammals and now I see why observations you can do you can observe apes but you don't have a controlled test with them because we had some tests also so there is a high resolution extremely sensitive video camera so that it would see in the night and this is in fact this is in my mind this is the most important goal just to see how that is where we followed it for ten month how hierarchy emerges in a society we live in a, of course, obviously in a highly hierarchical society on every level among physicists at the university in the political life it's one of the most common features of the society that it is hierarchical so we made a lot of movies with this arrangement and I will show you some this is again, as before, real time during daylight conditions so this is when they are sleepy, slow and everything was automatized for ten months so even feeding and giving water and everything was arranged through driven by a computer and with some mechanical devices here you will see this will be a little bit faster because it is now accelerated I don't know here are the days so this is now day, night you may have noticed that their behavior became much more excited of course accelerated but there are two rats in this thread wheel and these two ones I'm not sure, I cannot tell it's very difficult to stop it at the right time but they had a lot of fights so just watch what you see and they had injuries and everything it was a society with friends, enemies, dominance, subordinates and everything and of course we had the track of all of this with precision of a few centimeters because they were individually labeled by three colors so this is a very sophisticated program which was developed by one of the co-authors which eventually ok, I stop here this is red, red, blue, purple you can see that I hope this red is orange, blue, purple and this is red, green, blue so each rat had its color, bar this is a very long story it's very difficult to make them identify for a long time and I think this is the only ok, this is the first result I would like to show you this shows, you know, they go over the place and they turn up at a given time at a given position and there is an average for the seven members of the group and then you take one rat number one it's arbitrary and you watch how much time it spent compared to the average if it spent more time, it's red if it spent less time, it's blue and you see this particular rat liked to be everywhere this means it was on the top of the nest of the house much more than others and he didn't really like the thread wheel you know, the thread wheel it's not for rats they love it to do it and this is another rat and this rat spent most of his time in the thread wheel and you know, it's not trivial which one is the dominant because they like the thread wheel and prior studies concluded that the one who spends more time in the wheel is more dominant because that's the preferred position but it's not like that if a dominant male approaches a thread wheel and there is a pure little rat in the thread wheel he starts to rotate the wheel and the dominant male cannot get access to him so he defends himself by being in this preferred place now, I don't want to take much time every time this thing is written here here we destroyed we produced the passage among the four corners and this is the trajectories the heat mapping of the frequency occurrence for a particular for rat 3 and again, you know, it is why? it didn't like this one it liked this one the most it went to drink here instead of here this is not too much about the evolution of hierarchy but it is much more about the complexity of their behavior and we will see maybe at some point whether it is similar to how people behave I think they are similar it is but we could also find the dominance hierarchy as well because if you imagine that rat A dominates rat B then B most of the time runs in front of rat A and from this simple rule it is possible to construct a hierarchical network of interactions there is someone who is chased by everyone and there is someone who is not chased by anyone he is on the top ok, this is the last part almost within 40 minutes this will be already about robots but I started showing you that life ordinary life, you know, our life and robots is related, of course and this chimpanzee is apparently a curious one and interacts with the robot and the robot loses and look how curious this chimpanzee is ok, so this is a paper which was accepted 2 weeks ago into science robotics which is the science journal but it does not have an impact factor yet it is a new release by the science publishing group and it is about optimized locking of autonomous drones in confined environment and I don't have time to explain everything what I was going to but at least I would like to discuss just provoke your thinking that I would like to introduce this as a principle instead of, you know, some people are using this but I would like to claim that a good model of a complex a very complex system like drones are very complex, you know 30 drones are very complex there are time delays maybe it will be later shown there are a lot of effects which have to be taken into account so there are only a few rules but they have many parameters the rules themselves in a realistic case not in a simplest possible model so you just add one more rule which says so that you are entitled to use parameters in your model which are too many and the physicists, I didn't used to like and I still don't like models with too many free parameters because which one is relevant no, if you have a a bit fitness function so you know what you want your system to do you know, fly together or chase someone or explore an area you build a fitness function it's like the order parameter in statistical mechanics so you need a good function which says how well your system operates and then you optimize the parameters so that the fitness function had its maximum value and suddenly you get rid of maybe 15 free parameters out of 18 because for this you need to change a little bit about what you think about modeling you have to take the Darwinian thinking very seriously and imagine that anything which works very well has optimal parameters already so you don't have to guess you just optimize the evolutionary optimization that we used is not the most trivial optimization and so these are our drones there are 30 of them at this point one of the organizers had a fixed wing cluster swarm of drones these are very different cases I believe this gave us more freedom to put a lot of electronics onto the drone so they had a brain there was a Linux computer on each one and they had high very sensitive sensors and all of this was needed because perturbations I don't know how to come out from this so time delays is extremely I don't have to explain this to these robotics people but biologists may not appreciate how much trouble you get because of time delays your system starts to do this because it notices that they are too close only too late so then they are apart it's a long story if the arena is rectangular then they tend to get stuck into one of the corners and your model has to make them able to avoid all of this so this is a I would call this a simple but each term here has very natural frictional interaction with the wall interaction with the obstacle but each has parameters then we evolutionally optimize them and then for each velocity we want them to move in average with a given velocity so this is the parameter we would like to keep free so we feed into the computer and define for each parameter value over average velocity a set of values after optimization for the interesting parameters which are there which are important but they just come out as a result from optimization so this is how computer simulation looks like doesn't seem to be extremely sophisticated but you really have to enter this field to include all of the noises and the time delays to appreciate it what you see here looks like a computer simulation but it's not a computer simulation it is the track logs of an actual experiment of the drones they were released and this is how they moved around and it is very much like the simulation and it is also quite realistic I believe I am very close to the end I think this is the last this shows how 30 I think 30 drones in the field this is the lights of Budapest we are about 30 kilometers from Budapest over a broad area these are my colleagues here and they will chase this one and look how smooth this is the main point they are moving very smoothly in a very organized manner and at the same time they do the task we went to the field we had some long exposure photos of the drones and those sorts of funny pictures came out from this