Added: 4 years ago
From: tanevivan
Views: 37,003
Sort by time | Sort by thread (beta)

Link to this comment:

Share to:
see all

All Comments (77)

Sign In or Sign Up now to post a comment!
  • It's not evolving. It's adapting to its surroundings through trial and error. Not reforming its physical appearance to overcome the obstacle.

  • I challenge you to come up with an initial snake + fitness criteria + environment that will evolve a snake that props itself up, and rolls forward like a bicycle tire.

  • The only proof in evolution is through pixels, how convincing.

  • @darkaus32 haha, and it was a complex form, initially, programmed by an intelligent mind with pre set surroundings..

  • @bogan182 evolution is only a theory

  • @darkaus32 Why do you say "only"? A falsifiable theory which has not been disproven for several decades is the best you can get... *sigh* If people would just get the basic principles of science...

  • @darkaus32 This is NOT meant to proof evolution, it is just a simulation. Stop trolling your religious ideas, please.

  • @InfiniteRedPill Think of it this way: we know enough about evolution to simulate it.

  • @darkaus32 better to say the only evidence you accept is through pixels.

  • @escapeartist80 bones of humans and horse bones and chariots are coincidentally in the red sea, all of the bible record is true, rules etc, google red sea crossing, jesus is almost here and you still deny him haha, more disasters more problems in the world is his wrath, he will rapture his saints soon.

  • @darkaus32 And where did you hear all this? the insane asylum?

  • It's so real, it's disgusting to me!

  • oh...the sidewinder is awesome!!!!

  • That was fucking great!!

    One question: why our snake choose the best way to forward locomotion using rotation? the real snake always stays on top))

  • So is this a robot with preprogrammed movements, or does it sense and have a neural net or something. Every video looks like the snake is moving is some unchanging pattern...I'll bet if you expanded on the snakes senses and "brain" you could start making them fight each other.

  • This was really cool! keep up the great work

  • Good job, Ivan

  • I would like to know about the program that you use to make this simulation, I think it is very interesting, I like your job, thanks

  • Thanks for commenting and for your appreciations! The software consist of the (i) physical model of the snake robot and its controller, (ii) evolutionary subsystem to evolve its locomotion, (iii) realistic simulation of all the forces, and (iv) 3D visualization. I developed the (i) and (ii), used ODE by R.Smith for (iii), and OpenGL for (iv). For more details, please check below for my first responses to this video, and let me know if you have any questions about the implementation details.

  • no honey

  • Yes, but learned behavior IS a result of evolution, and the phenomenon is called evolutionary (or phylogenetic) learning. Other mechanisms of learning (of either a morphology or behavior, or both) include ontogenetic learning, cultural learning, epigenetic learning, etc. These are both observed in the Nature and implemented in AI.

  • Yes, and this type of learning is called evolutionary learning.

  • newb

  • jafo

  • Oh, and nice vid, btw.

  • By evolving?

  • Once computers are powerful enough, they'll be responsible for our evolutionary path.

  • We are already co-evolving with the technology we have created, and have been for some time. As the technology itself evolves its role in our evolution becomes exponentially more dominant.

  • You should read Ray Kurzweil's "The Singularity is Near" if you haven't already.

  • Look up framsticks. It's worth a play.

  • Yes! Framsticks! I bought the registration for that app. Very cool.

  • could be possible that we are just some kind of huge experiment also

  • Someone has. It's called Spore. :)

  • One day... one day

    (That game is still unfinished)

  • I know it is, but for the most part, it's done. Should be out soon enough.

  • spore has to do with inteligent design, not evolution

  • Oh for the love of... XD I should've expected someone to leave a comment like that.

  • sorry :(

    but spore istn't really the game that thelleht described, and i thought that was an important point

  • closest one that's ever been made, though.

  • i disagre, look up for "gene pool", its much simplr than spore, and a litle restrictive, but is true evolution in action, you strt out wit a bunch of incomptent swimers that eat and reproduce,if they don't eat they die,ther ofspring share the traits of both parents..you let the program run for a night and in the morning you will notice that your swimmers hve developped an almost homogeneous population of very succesful swimers, not realy a game, it leaves you with very little interaction, though

  • The creator of Spore believes in evolution, but the game certainly has some "intelligent design" aspects to it.

  • Yeah, it does have ID aspects behind it, but really, how fun would the game be if it was a 5-hour long movie and all you could do was watch?

  • If Spore was about evolution, it would be more like a computer-generated movie than a game.

  • Yeah, Spore was such a joke. It closer to a creationism game than it was an evolution game lol.

  • @kakkoister It was strange that it was about evolution but you are playing as God to change the creatures willingly.

  • @iUDods Dude, nobody can "play God": evolution (the natural selection of random mutations) eliminates the need for a God hypothesis. Even those who nevertheless believe one or more of the currently popular god-claims know better than to think in terms of "playing God." They acknowledge that evolution teaches NATURAL selection of RANDOM mutations. Natural means no gods; random also means no gods. Nobody is "playing God" -- not even God (that is, none of the objects of current theistic faith).

  • @cliffwalkinfool I was talking about Spore, dude.

  • Very interesting. Did you get to do a simulation with snakes moving through rough terrain or under boxes from the earliest stages or add predators the snakes would have to evade? I would like to see how likely it is that characteristics already found in nature would the ones best suited for animals' current environments.

  • Thanks for watching and commenting. Interesting questions, related to the emergent behavior, adaptation, ALife, etc... Yes, first I trained the snake in a challenging terrain with assorted obstacles (3'00" into the video) in order to obtain some general, robust gaits. The training was actually an evolutionary adaptation of fast (sidewinding) gaits, obtained on smooth terrain. Then I tested these robust gaits to unanticipated obstacles like pen, pile of boxes and burial under boxes.

    (continues)

  • Some of the emergent properties, similar to those in Nature:

    (i) Sidewinding locomotion, similar to Cerastes cerastes (Northern Africa) and Crotalus Cerastes (North America). The similarity between the snakebot and the real snake (which is the reason for sidewinding to emerge) is the uniform friction between he body and the surface (only uniform friction is simulated in the snakebot, and the scales, as a source of unidirectional friction don't work on sand for real snakes)

    (continues)

  • (ii) Increased elevation of the body of the robust gaits, and

    (iii) Increased winding angle of the robust gaits.

    These properties are not explicitly engineered by the creator; rather the creator describes what should be achieved (i.e. to move fast over obstacles) and the properties emerge as know-how obtained as a by product of the interactions between the problem solver (the evolutionary software), the snakebot and the simulated environment.

  • awesome

  • Well said.

  • Thanks for this. Truly amazing!

  • The morphology does not evolve?

  • No, the number of segments (15), the type of joints (universal), the number of actuators per joint (2), and their properties (max torque, max velocity, max turning angle) are fixed. Currently, I am working on the coevolution of the morphology of active sensing (orientation, range and motion patterns of sensors) and the locomotion gaits in constrained environments, with some promising results.

  • That's awesome! Keep up your awesome job. Thumbs up (y) Awesome.

  • niiiiiiiiiiiiiice did u make the program yourself?

  • Yes, I developed the evolutionary framework (the Snakebot is just one of its several applications). The physics is realistically simulated thanks to the Open Dynamics Engine (ODE). The 3D visualization is in OpenGL. You can see a snapshot of the software in the web page, indicated in the description of this video.

  • Thanks for this stimulation. It's good to know that I came from a retarded, mutated, orange snake.

  • you missed the point.

  • you missed the joke lol

  • And you missed the joke.

  • i think its good to know that i werent made from dirt after all :)

  • Nice work on emergent behavior. Most people commenting here don´t really understand what it is all about. Did you developed the program?

  • Thanks for commenting. Yes, I developed the software which includes the following components:

    (i) Evolutionary framework- evolution of the optimal motion patterns of snake's segments;

    (ii) Realistic simulation of the physics (thanks to the Open Dynamic Engine by R. Smith), and

    (iii) 3D visualization.

    Some publications are listed in my web page (URL is in the description of this video). Please e-mail me if you can't access a paper - I will send the PDF.

  • Nice work. How many generations are you needing before emergent behaviour is useful?

  • From 14 to 35 generations. The variations are due to the element of randomness in the creation of initial population, selection (binary tournament) and genetic operations (crossover and mutation). One generation may require from 12 to 15 min of runtime. The population size is 200.

  • Aww nvm, I see what you mean, with respect to the head. Got ya

  • Umm the sidewinder is a species of snake that moves sideways.

  • No it doesnt many snakes (especially in the desert) move like that

  • i see. i didn't spend too much time reviewing the video, but i see now that it's goal was to simulate sidewining, in which case i suppose this could be called a success.

  • Actually, I believe it was just going for any motion, and the sidewinder came out.

  • Umm actually there are several species of snake that move sideways, educate yourself.

  • yes ill get right on that: I will make sure I know all the possible movements of all possible species of the world by tomorrow. Thanks for turning my life around, asshole

  • Good evolutionary simulations would be extremely complicated. To create a simulated environment in which for example the tiger's bones become stronger because of mutation would require vast programming knowledge and computational power or simplifications that make simulation more strict and rigid.

  • To work around tedious programming issues one could just set a number of natural laws, start up the simulation and let the system work according to these laws (just like our universe). But that would require tremendous computational power.

  • Ah, but that's been done.

    Check out Chris Adami's work with the Avida artifical life software package.

    The work his team did actually showed how complex features can arise via the laws of natural selection, by computer simulation as you suggested.

  • Great Work, i love stuff like this

  • very nice

Loading...
0 / 00Unsaved Playlist Return to active list
    1. Your queue is empty. Add videos to your queue using this button:
      or sign in to load a different list.
    Loading...Loading...Saving...
    • Clear all videos from this list
    • Learn more