Game Idea: Real Time Strategy

[angeladik]

Hello everyone

As mentioned im my "Hello..." post, I'd like to make an effective RTS bot for player educational purposes. Of course battle between AIs may be a great experience too - we had an opportunity to watch Ants in action. The idea is simple, how about creating a simple RTS as a field of battle for our bots? Difference in units would increase possible strategies (one unit would be a counter for another, like ranger vs infantry, cavalry vs archers and infantry vs cavalry). Different resources would define gathering aproach and differ it depending on strategy of bot. Such game would require both military and economical approach. By applying little bit of graphics we could watch really interesting games between AI.

This is just a simple example of what could be done. In addition to this bots could learn over time (neural network), evolve (genetic algorithms) and even mutate to create unexpected (but maybe effective) parts of a good bot. There are many more things which could be included so I invite everyone to talk about this. Go ahead and tell me what you think about it.

Unfortunatelly I am not experienced in coding so far, so I ask you, experienced coders for feedback, how complicated would it be? Are we able to do this in required time, or is it much more complicated than I think?

All requests are welcome, no matter if you are experienced coder or just a newbie like I am. I am really excited about coding and AI and hope that it will result in bots. Hope you share my point of view.

Adrian

P.S.

If you are looking for more ideas, just ask me, I have plenty of them

[Parasprites]

A) Real time games aren't compatible with the way the contest works. You'd have to make it turn based
B) This is basically just a more complicated version of Ants already.

[MagnumPU]

I definitely think this is a good game idea, it might not be a good next game idea though. I recently posted that I thought it might take 8-12 months before the next game, based on the amount of time between the last two games. Someone responded and corrected me, explaining that the organizers spent a lot of time setting up ants so that a lot of the contest code could be reused in future contests. The real time idea that you have I think would mean another re-write, so it's a good idea, but probably not for the next contest, unless we want to wait a while.

[angeladik]

Parasprites, thank you for fast response.

You said that game should be turn based. Is it possible to make it turn based? Actions taken by AI, like chopping wood, mining, fighting etc. would take required ammount of turns, like 1 turn for getting 1 unit of material, 1 turn to move unit, etc. Of course I can be mistaken, because looking at Ants the maximum ammount of turns was 1 000. It may be not enough for Strategy Game like I am thinking about.

About similarity. All previous contests were somehow connected to the battle. I think this one would not only be similar, but also more challenging. Making battle units different would require more than comparison in number of units, this may lead to interesting results. In addition, battle brings more emmotions.

As I mentioned, please point all weak points of this concept. I'm a beginner with AI/programming so every problem may be a good one to talk about. Correct me in any point which requires it. The more mistakes you will reveal, the faster I will learn. Please make a clarification why it is a problem and if you can, tell something about solving it.

Thanks for your reply
Please keep commenting with all suggestions, ideas etc.

Adrian

[angeladik]

Magnum

Thanks for post. I made this post just to give a topic to talk about. I appreciate your effort in getting info about the process needed to make a contest. Time isn't that important for me. I want to test myself as an AI/program creator. The longer I will wait, the more ideas I will have and the bigger knowledge about AI/coding I can obtain.

All posts are welcome

[zarkon]

[MagnumPU]

[Vogons]

Interesting, what bots are using this technique ?
I know i could have looked at the sources myself, but my 2nd son decided otherwise and was born the day after the submissions were closed

François (ThisIsNotABug)

[Vogons]

Forget it, just found the answer in the forums (Xathis & BenJackson).

[angeladik]

Congratulations Vogons on the occasion of birth of the son!

Thank you for a link MagnumPU, now I will read this and I will let you know what I think about it. I've recently read a book about AI, it was focused on neural networks and genetic algorithms so I still have them in my mind and I think where they could be used. Of course my experience with AI and coding is low, so I look for any new sources, e-books, books in my School library etc. All additional links are welcome.

I am glad that you share my point of view zarkon. That is what I was thinking about. Short turns may seem for people like a fluent action, while AI works faster and have a turn based game instead. Of course there lies advantage of AI, which is able to make more moves, but that is what are we looking for, really good AI game. My oppinion is that we need to give bots a good field where they can proove their worth. Of course the game should not be too complicated, because it may result in too long calculating what bot should do (not mentioning about the length of code). But making a simplified Strategy Game, more complex even in few aspects than ants, will bring great experience and will multiply possible situations.

Looking at similarity between Strategy Game and Ants, for AI it doesn't really matter what the graphic representation is. I think that all previous projects (Ants, Tron and Planet Wars) had similar parts. The bots job is always the same, win. I don't think that my idea is best, it's just one them. I will soon post another idea(s) and we'll have more topics to talk about.

Thank you all for response, keep doing it.

Adrian