My 2D graphics skills are poor, and 3D is virtually non-existent. Can anyone recommend a console mode app that would be a fun/interesting AI challenge? [My site|SGI STL|Bjarne FAQ|C++ FAQ Lite|MSDN|Jargon]
Ripped off from various people

Chat program...
Any number of board games you could do with a text interface (as long as the board is not too big)...

Ro-Sham-Bo (rock-paper-scissors).

If you can get an AI to do better than 50% wins, then that''s interesting.

Timkin

simple board games like noughts and crosses (tic tac toe) and connect 4 work well with text interfaces and are fairly simple to do (and as such are good for learning with)

How about displaying the user a grid showing troop positions - have various obstacles and/or terrain types. Then you could go into looking at squad based tactics, controlling choke points, resource management and all sorts.

Just make a simple rule-set for the game like only pieces can be moved per turn. That would be interesting - the AI would have to weigh up which pieces it would prefer to move to gain the better advantage even if it means allowing the human to attack and kill some of its other pieces.

You could even go as far as to make 2d line of sight algorithms based on what terrain the units are in, and what obstacles are in their way. All this off a simple grid.



regards,

GeniX

www.cryo-genix.net

quote: Original post by GeniX
How about displaying the user a grid showing troop positions - have various obstacles and/or terrain types. Then you could go into looking at squad based tactics, controlling choke points, resource management and all sorts.

Just make a simple rule-set for the game like only pieces can be moved per turn. That would be interesting - the AI would have to weigh up which pieces it would prefer to move to gain the better advantage even if it means allowing the human to attack and kill some of its other pieces.

You could even go as far as to make 2d line of sight algorithms based on what terrain the units are in, and what obstacles are in their way. All this off a simple grid.



regards,

GeniX

www.cryo-genix.net

Wow, that could be cool... I imagine the AI would be very similiar to chess. The line of site is good idea too, I can see snipers and artillery being put to good use.



[My site|SGI STL|Bjarne FAQ|C++ FAQ Lite|MSDN|Jargon]
Ripped off from various people

yeah.. best of all is that with a days coding you can get the interface and system in place using only text-mode - just represent units with ascii characters.

and you''re all set to focus on the AI!


regards,

GeniX

www.cryo-genix.net

quote: Original post by wild_pointer
Wow, that could be cool... I imagine the AI would be very similiar to chess. The line of site is good idea too, I can see snipers and artillery being put to good use.

I find it amusing, interesting and I guess a little disheartening that the only AI topics that ''appear'' interesting to the majority of people are those that involve virtual characters running around shooting things! This is not to say there isn''t an interesting AI challenge in there, but believe me... if you could solve Ro-Sham-Bo at better than 50% you will have achieved something, learned something and found something useful!

Learn to crawl before you try running!

Regards,

Timkin

"I find it amusing, interesting and I guess a little disheartening that the only AI topics that ''appear'' interesting to the majority of people are those that involve virtual characters running around shooting things! "

Virtual characters provide an excellent basis for investigating the relationships between distinct agents in a world, and also looking at algorithms which are current and in use by modern games. From Line Of Sight, Path Finding, Squad based Movements, to even learning to predict player movements.

Making these virtual characters fight or shoot seems a logical next step as it manages to provide some added benefits:
1) The AI now has an intermediate or secondary purpose between its introduction into the world, and winning the game (like defending the top of the mountain successfully)
2) The AI is now applicable to a large majority of games out there. This is useful to understand whats being used in the industry, and also that the large useage thereof means that the techniques are well documented - even on a mostly free medium such as the internet.
3) The AI is now becomes Game AI, which is both relevant to the posters interest, and introduces the coder to current issues with game ai (should the computer be omniscient? or have the same restrictions as the player? what about advantages only for the computer players?)

While I do agree with your statement that there are a number of topics that can be investigated, and it doesnt *always* have to be characters running around fighting, it is an area of interest for most game developers. They can both relate to having thought a lot about such strategy games, and having played quite a number of them.


"Learn to crawl before you try running!"

Point taken. I guess I was making assumptions on the posters level of experience, and what he/she wanted to look into. Then again, it was merely a suggestion..





regards,

GeniX

www.cryo-genix.net

ok the rock scissors paper thing is cool.

at first i was like that is silly the human could always go random ... i mean pure random with dice. and all the computers "thought" would be for not.

cept if we gave the human a motivation to win. then the human would be trying to predict the computer. taking advantage of the computers flaws.

yet thats exactly when the computer gets room to predict the human: when the human is trying to predict the computer.


in fact i think our first insight is this:
the computers success in predicting will be porportional to the humans success in predicting.

in other words the computer would have to toss the human a bone. exactly what a human would try to do to a computer. establish a pattern then manipulate it.

all of this reminds me of how i used to play poker. you see i grew up playing cards for my allowance. my dad just went right along. wasnt till i was bout 12 that i had any chance of not loosing my allowance to my dad(this started when i was 6). anyway the thing i gleened from my misspent childhood was that you have to be predictable to control your opponent. if you are completely unpredictable in your dealings in poker then you have zero control over how your opponent bets. which is fine if you want to break even. if you want to win however you got to have some control of how the bets go down and that means being predicatable.