Hello there have probobly been some around before dreaming and imagening stuff about this or maby have made some tries on it. If you know any good info or threads please link em. I since i am a game developer(programmer) and want to continue educating my self in the areas of my intrests i am now beginning from what i can see a lifejourny developing a good base for virtual world. This project will not be opensource from the start byt might be so later on. It has been a small dream i have been building up over the past 10 years of how i want a "mmo" to be. And no i am not going to try to build a mmo at this time. What i am looking to build up is a multiplayer virtual world with undefined numer of players mainly focusing on the world and the evolvement of the world. So why am i posting here? I have some idea of how to integreat the spicies controlles and dependensies. But thay might not be the best. I want the content of the world to be generated depending on the current sitation / naboring spicies and such. I dont plan to generate the vegitation but more just placing premade vegitation for players to be able to recogise. As well as creatures and predators. And in all this mess the player race will start developing. So the big question is: IS there any good already known ai patterns to help with this? The full project will involve stuff like evolving world were spicies and vegitation move around or just dissapear becouse conditions are not right. as well as some stuff will become reall rear becouse conditions for it to be spawned will be hard to fullfill. All this will allow the players in some fasion controll the world and make an impact on it. I am not sure if i want to involve weather or not. Might be hard to simulate weather in a rather small sandbox enviroment. but as i said any help or pointers will be helpfull. Also if this projekt is of intrest to more people i might consider making some larger publication of it but not as it is atm becouse we are just researching possibilites and such. Best regards Athos

It is a very lofty goal that you want to achieve. I can not really suggest any good links to such a massive project as what you want to do is so hugely complicated that you would be virtully on your own in all aspects of its development.

Before I offer pointers in the field of "AI", I must ask what your level of experience is with programming. If you do not have a very high level of programming experience (at least 5+ years) then I would highly recommend focusing on more manageable project first. You should never lose sight of your goals, but I can tell you right now that without the experience this project will quickly and easily become unmanageable.

That being said I am not quite sure how you want the interactions to take place... One possible place to start is by researching Predator-Prey relationships: Predator Prey Simulation Plant Population Dynamics

You see what you want to do seems less based on "AI" and more based on growth rates and other mathematical simulations.

"You see what you want to do seems less based on "AI" and more based on growth rates and other mathematical simulations."

Hehe both agree and dissagree with you on this point since AI is a very loose term as it is. I would consider predetor pray system and dynamic growth system as you suggested as AI. Also since game actoers behavoir is a big intreste for me i intend to research it more.

I have more then 5+ years of programming behind me. Aswell as what i posted. And i currently work in the gamedev industri as a programmer so hopefully i have sufficant skills to take on this project allthough i dont intend to rush anything but to lern maby write a few papers and make a fun project with my girlfriend =)

so somthing aloing the lins of high fantasy worldsimulation with strategy elemetnts and such.

Predetor pray system is more a simle funtion to controll the ammount of creatures to spawn each spawntick. More food more spawn less food less spawn. Also useed to determin were the predetors will spawn and live( close to pray but as far away from its own predetors as possible ) or somthing like that.

i Will try to conjure a simpel basic no heavy graphics example in xna to simulate somthing like this.

It really all just depends on how complicated you want to make things. I am not sure if you are going for semi realism or you just want to code some crude interactions to make the environment appear slightly less static.

Of course for creatures you could (and should probally) use state machines... I am fairly sure you already know of them but here is a good tutorial if you don't: State Driven Agent Design

Plus I am kind of a fan of steering behaviors: Steering Behaviors For Autonomous Characters

I am not going to suggest neural networks just because I feel they are a "buzz word" and don't really have a use in this situation...

i agree with you. on the neural nets. since before i learned about them and used one or two small nets i feelt like you coudt do probobly anything with nerutal nets and you can probobly do. But it is kinda hard to find good implementation of it. And i dont really see the use for it in larger games.


I want somthing more then crude implementations but still one has to think of the preformance and cant really spend to to much on small stuff. Since what i am thinking has "AI" on so many levels for controlling the world. So i think it will be heavy for the cpu anyway.

My game idea is simply to fill a virtual world with zombies and give the player a procedurally generated city to run around in, shops to loot, and lots of things lying around to hit the zombies with. I've been working on it for two years, learning lots of stuff as I went along. How long will it take you do do what you're after? For me, game development is just a hobby, which I've been balancing with my official education. 1200 odd noob posts later I've learned a hell of a lot, but my dream of a zombie physics sandbox game are still out of reach and I've not even thought about multiplayer yet.

Short version: Start with something small and build it up, rather than aiming for something big and jumping over too many hurdles.