AI Contest Type Thing

Pages: 123
Bumping, maybe we can actually do an entire competition, once.
@ SGH: Like Disch said, we need a common framework to start from. I think the biggest problems with the competitions we form here are either:

1) The Poster who suggests it isn't experienced enough to follow through with the project.

2) We as a community hijack the idea from its progenitor and change so much of it that it isn't even their idea any more so no one ends up building the ground work.

Basically we need someone to take the lead and build an idea BEFORE they present it here. But I even remember a few cases where that failed to take off.
Yeah that's why I'm quickly whipping up a header file that implements a bit of a sketchy ascii board with the idea other people can create their AIs and release them as header files and then they can be played against each other using this little ascii. It just accepts moves, checks legality and makes them. I figured although the implementation is likely to be a little questionable, as long as it works, it will do and no-one else seems likely to do it at all.

I should really have finished already but I got a cold and am being unmotivated and lazy because of it. =/

@Disch - This is what my related question was about namespaces/headers.
Last edited on
@Mats: I thought the same.
In fact I did an OpenGL 2d engine which is working pretty well...
https://dl.dropboxusercontent.com/u/83943521/Screens/f/cplpl/quoridor.png

But I am waaaay to lazy to set up the rules.

BTW the bump was for not to get the topic lost.
I have the rules almost completed. If I sent you my move validation function is that helpful?
I'd call that teamwork.
But i'm sure you'll have troubles validating where a wall could be placed at, even if I can give an eye on that (thanks to my pathfinding algorithm I've done a while ago).
I was thinking about valid walls today too. After thinking about A-star a little bit, I realized that the framework does not necessarily need to create paths to figure out validity. Pathfinding is probably the most straight forward, but another way might be to make a list of valid wall placements.

It would be up to the AI to figure out the best path ( which probably should lean towards options rather than distance ).
@Lowest: No matter what the AI does, it's part of the game logics to find a path from both pawn to the enemy "spawn" and see if they can reach.

We must validate user input, too, that's what I mean, or players will just block them off and win right away.
it's part of the game logics to find a path from both pawn to the enemy "spawn" and see if they can reach.


Why? What does the game care what path the pawn takes? It only needs to validate that there is a possible route. I do think that a good answer is to make a path, but it isn't the only answer. Nor would the path have to be the best path.

For example, an algorithm like Astar. Part of Astar is going back upon the built path to make it better. This is not a necessary thing for the game to do.
Topic archived. No new replies allowed.
Pages: 123