evolution simulation

Does anyone have or know of a good evolution simulation program? I am thinking of generating my own program.
1. create environment (gravity, temperature, food sources, etc..)
2. introduce primordial life form(s) to the environment.
3. Randomly mutate individuals.

I have often thought of an AI program that can allow for these evolving creatures to learn how to use tools and solve problems. For example, a round 2D organism with one arm of length x eats by rotating it's arm to catch food. Now stick segments can be added to the environment. Can the organism learn to pick up the stick and use it to catch food which is further away? Can it learn to use a number of sticks to create a basket to collect and store food? Can it teach its friends? Can they form groups? .. etc...etc..
Obviously, for this to approach real AI only the most basic features must be initially programmed into the creatures like curiosity, memory, need to survive and procreate.
And wouldn't it be great if the virtual world allowed us (God) to modify the environment or introduce a new species?

There must be something like this out there.

 

Log in or Sign up to hide all adverts.

Here are a couple of programs w/source:

From the book - The Computational Beauty of Nature:

http://mitpress.mit.edu/books/FLAOH/cbnhtml/home.html

Some small bits and pieces to get you started.

These two pages have programs of varying degrees of complexity, but it should be enough to give you a couple ideas:

http://dmoz.org/Computers/Artificial_Life/Artificial_Worlds/

http://www.ai.mit.edu/people/unamay/trellix/ed-mar20/id5_m.htm

Also, I've read some really good papers at:

http://citeseer.nj.nec.com/cs

No source, but some nice ideas.

 

Log in or Sign up to hide all adverts.

dsdsds,
You are not gonna get very far with this approach. Your hardware (as well as your brain) is not powerful enough to truly simulate complex worlds which include intelligent creatures which have (for example) functional imagination and memory. It takes lots of resources to support just a single artificial brain-like system. That's why our brains usually use relatively complex if-then logic instead real simulations. You have no way to really imagine all powers which are moving things around you and running our world. You just cannot go global with all important details.. You need to switch direction. You have to use local thinking and keep expanding it based on available resources. Whenever you can, generalize, remember rules and forget details. That's the key.. Key which you cannot effectively use when running the global simulation..

 

Log in or Sign up to hide all adverts.

dsdsds,
Have you read any of Dawkins's writings on this subject? He's inspired some pretty basic simulations, of which the best I know is Netspinner.

Unofficial Dawkins homepage: www.world-of-dawkins.com

 

This one is nice

Please Register or Log in to view the hidden image!

 

The technique you're describing is called evolutionary computation. (also, genetic algorithm, genetic programming) There are many, many systems out there that use this concept. Begin my usual list of links:
Hitchiker's Guide to Evolutionary Computation
ENCORE
Amygdala
Zooland has a long, long list of ALife resources
One worth taking a look at is Framsticks

The general AI sites often have some introductory materials about EC/GA/GP:
generation5
ai-depot
ai-junkie

Good luck! Genetic programming has become one of my passions, so I'd love to hear your thoughts.

 

I like lifedrops and: http://www.virtual-worlds.net/lifedrop/
genepool http://www.ventrella.com/

 

Thanks for the info and links everyone. I know it's not as easy as my post seems to suggest. I'll start reading.

 

It's also not as hard as it seems.

Please Register or Log in to view the hidden image!