Hexapod walk

[Marmakoide]

Here it is two videos of a hexapod robot that walk :
http://www.lri.fr/~devert/videos

Physic engine = Newton Game Dynamics, of course
Graphic Engine = Ogre 3d
Controler = Dynamic system, limit cycle optimised by an evolutionary algorithm
Evolutionary algorithm framework = Open Beagle

Walking gaits optimisation time (Linux 2.6, Pentium 4 3 Ghz) :
Walk -> 0H15
Run -> 1H30

Thanks for all the help from Julio.

[aidave2]

is it using some new codec?
videos wont play on my system

[Marmakoide]

I used mencoder to build the video, with the following command line :

Code: Select all

mencoder mf://*.png -mf w=256:h=256:fps=25:type=png -ovc lavc -lavcopts vcodec=mpeg4 -oac copy -o araknid-28-04-2006.avi


The videos are played without problems at least with mplayer, xine and totem, within Linux (no Windows machine under the hand...)

[kallaspriit]

Tried with 4 different programs.. you should consider using something else to make the video..

[Gianluca]

I don't have problems with videos!!
It is a interesting exapod

kallaspriit ... try with a different O.S. (e.g. Linux)

[Marmakoide]

Tried with 4 different programs.. you should consider using something else to make the video..

? I believe I use a Divix or Vidix encoding... Which codec I could use, to make available for my videos for Windows ?

[Julio Jerez]

I can not play it either, I even tried to resample it with VirtualDub, but it sayd it is is using a unknown codec called FMP4

[Marmakoide]

mplayer said that about the codec :

Code: Select all

Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family
Selected video codec: [ffodivx] vfm:ffmpeg (FFmpeg MPEG-4)


I think I can translate this to another codec... But which one ?

[Robert Isele]

I couldn't play it either

I think I can translate this to another codec... But which one ?

i think MPEG2 is the most common one, but it produces bigger files than XVid and DivX.
So if size matters XVid is a better choice

[Julio Jerez]

I finally saw the videos.
For thoss tah want to see you can download this codex
http://www.free-codecs.com/FFDShow_download.htm

Man the run cycle is absolutely awesome; you are telling me that the genetic algorithm came out with that solution all by itself.
The run cycle really looks organic and that it came out from an evolution process. I guess that having a clear feeback really improve the evolution of the genne and reduces the mutations that lead to such moronics solution like I had seen.
I have seen many creatures walking with thist algorithm with other engines, but I never saw one with such symmetry and coordination.
Congratulation, I am looking for to see more.

I most say that the walk cycle is not as impresive as the run.
one question after how many iteration did you get that result?

[The unProfessional]

Careful with that codec... on my home machine it messed up playback of regular xvids.

[MiggyE]

That's really cool! It has subtlety of movement that makes it look alive. Nice work!

[Marmakoide]

I finally saw the videos.
For thoss tah want to see you can download this codex
http://www.free-codecs.com/FFDShow_download.htm

Man the run cycle is absolutely awesome; you are telling me that the genetic algorithm came out with that solution all by itself.
The run cycle really looks organic and that it came out from an evolution process. I guess that having a clear feeback really improve the evolution of the genne and reduces the mutations that lead to such moronics solution like I had seen.

Absolutly, using the good inputs (angle insteads of force) is less noisy, so it allows a better exploration of the solutions space. The other important point is the design of the controler : made from CPGs (Central Pattern Generator, see Auke Jan Ijspeert papers). Basicly, it's a network of non-linear oscillators. Those oscillators are a simplified model of a neuron group found in the spinal cord of the lamprey.

I have seen many creatures walking with thist algorithm with other engines, but I never saw one with such symmetry and coordination.
Congratulation, I am looking for to see more.

Impossible without a good physic engine...

I most say that the walk cycle is not as impresive as the run.
one question after how many iteration did you get that result?

Here it is the parameters of the evolution :
* Genetic algoritm : CMAES
* Population : 15 individuals
* Mu+Lambda ratio : 2
* Fitness measure : distance performed by the robot

Walk cycle :
* 30 seconds of evaluation per individual
* 5 generations
* 0h15 to evolve
Well, 5 generations, it's mostly random search

Running cycle :
* 90 seconds of evaluation per individual
* 30 generations
* 1h30 to evolve

Very soon : a snake and an articulated 24 legs bug...

[Julio Jerez]

The runner look like one of those bottom feeder creature you see on those PBS documentary.
It would be intersecting to see if obstacle could be added to the landscape, also something like claws for feeding. It may be pushing too much but it would be a good challenge to se how the neural net record that kind of input.

[Thomas]

wow, very cool! i am looking forward to the new creatures. (the vlc player (www.videolan.org) played the videos without the need for any codec.)

i am just curious. how far do such evolutionary simulations go currently? what would be possible with the world's fastest super computer for example? i find this very interesting.