Author Topic: Idea for the physics of bots environment.  (Read 5039 times)

Offline deoxymoron

  • Bot Neophyte
  • *
  • Posts: 15
    • View Profile
Idea for the physics of bots environment.
« on: April 25, 2010, 03:01:35 AM »
I have an idea for the physics of the bots environment.

instead of having qualities applied to the area, like the vicosity for example. perhaps instead this effect could occur if actually particles representing the medium which the bots live in could be used.

what i'm imagining is little tiny dots representing water/air/whatever particles as the medium.


The advantages:

*bots could develop way to manipulate the meduim to perform better. ei. swimming (perhaps with help from other bots)

*tiny tiny bots would be at a disadvantage as the medium would seem thicker for them- harder to push the dots out of the way.

*Also if random movement for the dots is used, tiny bots would find it hard to have a useful dna code to find food if they're constanly being pushed around. bigger bots won't have this problem.

*overall more accurate simulator of reality, esp if modeling bots on 'replicators' like RNA or just cells.


The Disadvantages:

*seems like a cpu hungry system to enact to program?

*would mean more complicated settings.


hope this hasn't been brought up already, if so disregard


good idea? discuss.

Offline bacillus

  • Bot Overlord
  • ****
  • Posts: 907
    • View Profile
Idea for the physics of bots environment.
« Reply #1 on: April 25, 2010, 04:15:30 AM »
Seems like a good idea if you have a supercomputer handy, but I think any ordinary computer would break down with all the calculations involved. Step one would be to get the core functions running with minimal calculations before anything that spectacular can happen.
As for the idea itself, I could see how it could help with some details such as drag etc., but otherwise I don't see how it would differ from just using simple viscosity simulations.
« Last Edit: April 25, 2010, 04:19:46 AM by bacillus »
"They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown."
- Carl Sagan

Offline ashton15

  • Bot Builder
  • **
  • Posts: 99
    • View Profile
Idea for the physics of bots environment.
« Reply #2 on: April 25, 2010, 05:16:40 AM »
Quote from: bacillus
Seems like a good idea if you have a supercomputer handy, but I think any ordinary computer would break down with all the calculations involved. Step one would be to get the core functions running with minimal calculations before anything that spectacular can happen.
As for the idea itself, I could see how it could help with some details such as drag etc., but otherwise I don't see how it would differ from just using simple viscosity simulations.

that said the falling game of sand which seems to be some kind of CA runs well enough, just things like water flowing acurately don't tend to work well (see algodoo and oe cake) so I suppose I agree with mr green tentacle monster really . Another thing is why can't the advantages you described be simulted easily by areas anyway? And if you were to employ particles like that I'm not entirely sure it would actually be realistic as most cells are many times larger than particles which I think you mean to represent at a moleculur level. I definitely think things need to flow though which might be slightly harder to do.

P.S: sorry to disagree with you the first time I spoke to you
« Last Edit: April 25, 2010, 05:17:36 AM by ashton15 »

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Idea for the physics of bots environment.
« Reply #3 on: April 25, 2010, 03:00:47 PM »
A proper fluid simulation is crazy expensive CPU wise.  There was a fluid simulation version of Pong floating around on gamedev.net a while back.  It brought my 2 ghz machine to its knees.

As ashton points out, using something like the falling sand game might work.  But it's still a lot of computation, and it doesn't necessarily work well combined with "real" physics.

Offline deoxymoron

  • Bot Neophyte
  • *
  • Posts: 15
    • View Profile
Idea for the physics of bots environment.
« Reply #4 on: April 25, 2010, 05:36:45 PM »
Hm, thought so... oh well. It seems like an important effect for evolution of multibots (something i'm trying to achieve)

is there any other (non-cpu hungry) system that could produce this effect. i mean, i really think the bots need a medium to be able to manipulate. i suppose it doesnt hav to have absolute real physics programmed into it, just something for the bots to push. this is how it would differ from normal vicosity simulator, i think? lol :/

I'm quite sure its currently not happening in this simulator? is it? i dont know. im still a dawinbots noob for now.

Nope, no supercomputer for me.  
currently using q6600, 4gb ram, 1gb video mem. win xp


Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Idea for the physics of bots environment.
« Reply #5 on: April 26, 2010, 12:54:23 PM »
Well, I do want something like ant nests.  So bots need to be able to dig in to shapes.  I can also see something like "wind" or "current" getting implemented.  Something like a static vector field you could import or paint in that could move materials and bots around the world to keep things from stagnating too bad.  And of course I'll figure out the physics to simulate things like squid bots (multibots that can move around using jet propulsion) and fish bots (undulating motion to produce motion) and swim bots (breast stroke motion to produce movement).

Offline deoxymoron

  • Bot Neophyte
  • *
  • Posts: 15
    • View Profile
Idea for the physics of bots environment.
« Reply #6 on: April 26, 2010, 05:46:43 PM »
cool, i look forward to seeing that!

Offline bacillus

  • Bot Overlord
  • ****
  • Posts: 907
    • View Profile
Idea for the physics of bots environment.
« Reply #7 on: April 27, 2010, 01:16:01 AM »
Quote from: ashton15
that said the falling game of sand which seems to be some kind of CA runs well enough, just things like water flowing acurately don't tend to work well (see algodoo and oe cake) so I suppose I agree with mr green tentacle monster really . Another thing is why can't the advantages you described be simulted easily by areas anyway? And if you were to employ particles like that I'm not entirely sure it would actually be realistic as most cells are many times larger than particles which I think you mean to represent at a moleculur level. I definitely think things need to flow though which might be slightly harder to do.

P.S: sorry to disagree with you the first time I spoke to you
I guess the falling sand game works fine, but I have a sneaking supspicion that the physics behind that is horribly oversimplified-it's something closer akin to Game Of Life, and mixing discrete and nondiscrete can be a bit of a nightmare  
Always good to get some new ideas flowing though  


Quote from: Numsgil
Well, I do want something like ant nests.  So bots need to be able to dig in to shapes.  I can also see something like "wind" or "current" getting implemented.  Something like a static vector field you could import or paint in that could move materials and bots around the world to keep things from stagnating too bad.  And of course I'll figure out the physics to simulate things like squid bots (multibots that can move around using jet propulsion) and fish bots (undulating motion to produce motion) and swim bots (breast stroke motion to produce movement).
Shape digging and propulsion are things I would really want to see-how much of the physics could be generalized into one variable with multiple instances, so you can iterate through instead of checking if Feature X is active? More generalization would free up some memory, especially when minimizing environmental features (No doubt you've thought about this already, just curious as to the extent   )
"They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown."
- Carl Sagan

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Idea for the physics of bots environment.
« Reply #8 on: April 27, 2010, 01:30:35 AM »
Quote from: bacillus
Shape digging and propulsion are things I would really want to see-how much of the physics could be generalized into one variable with multiple instances, so you can iterate through instead of checking if Feature X is active? More generalization would free up some memory, especially when minimizing environmental features (No doubt you've thought about this already, just curious as to the extent   )

The goal is to set things up so that:

a ) the physical dimensions of the simulation do not negatively impact performance.  That is, you only pay CPU for things that are actually in the world, and not for empty space.
b ) the simulation can run in linear time for the number of "features" in the world.  Like bots, shapes, etc.
c ) the memory cost of the world is linear to the number of features, and independent of world size (so same thing as a and b, but with memory).

Beyond that it's hard to get in to specifics.
« Last Edit: April 27, 2010, 01:31:51 AM by Numsgil »

Offline bacillus

  • Bot Overlord
  • ****
  • Posts: 907
    • View Profile
Idea for the physics of bots environment.
« Reply #9 on: April 27, 2010, 05:48:31 AM »
No need, that's all I wanted to know  
"They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown."
- Carl Sagan

Offline peterb

  • Bot Destroyer
  • ***
  • Posts: 148
    • View Profile
Idea for the physics of bots environment.
« Reply #10 on: April 29, 2010, 11:37:57 AM »
Ants digging.. hmmm


It could be done with a cell type which isnt algea but is rock (no feeding valeu) of solid color (maybbe a lot of .shell only)
Rock would be a bit hard to digg i mean kill (like requiring a special shot type), a shot type that would also have more effect on .shell
Rock would normally be fixed pos (however as some animals use rocks as to their protection.
In essence a solid rock needs a default fixed state (but once shot at a few times becomes free to move). (might  be attracted by gravity)
A smart animal would use this as shielding...  (or even collect it for shielding).

Oh and rocks dont duplicate, might split by shooting into smaller rocks.
But so far i've never seen a rock in an act of replicating.  
Also they dont have vision
So .. they are almost like a raw basic cell no memory regions no DNA execution(* perhaps for splitting a little command),
maybe some different rocktypes (hard harder, soft, seize, maybe squares or circles or hexagons, or simple triangles.).

Except for the non round shapes Could be done in current DB
But DB engine could be optimized for them in a next version, since they require far less  computation.



Offline ashton15

  • Bot Builder
  • **
  • Posts: 99
    • View Profile
Idea for the physics of bots environment.
« Reply #11 on: April 29, 2010, 02:22:36 PM »
Quote from: peterb
Might split by shooting into smaller rocks.
If implemented I'd say conglermerations of rock should be able to recombine (in DB3 there I heard phargmotis was bieng made which would probably be the best way to do it (eek can't spell)) thus you will not have many smaller rocks eating up processing speed