I wanna

First of all,

I'm a "long time reader, first time blogger", so I'd just like to say Hello to the GG community and anyone else out there! I've been surfing through the forums and resources at GG for some time, but haven't really had anything "blog-worthy" until now. : )

So what do I think is so blog-worthy? Well, the last couple of weeks I've been very intersted in game AI and how it's done. I picked up a book at Borders - "AI for Game Developers" - and was suprised and excited that you don't need to be rocket scientist to give your computer players a little decent AI. I'm happy with the coverage that the author gives all of the various subjects; Following/Evading, Path-finding, Finite State Machines, ect.. The book gives a nice overview of a subject without being intimidatingly specific, so I'll definately wan't to get some more books that are more-in-depth in certain areas, when I get to that point.

The first thing I wanted to tackle was path-finding. I went for the ever popular a* method, and decided that the best way to understand exactly how it works, is to just [u]do it[/u]. So far, I've got a klunky system worked out in script that (surprsingly!) works; I'm planning now on working out something better in code. But with the script version, I think that the hardest part was concocting a function that automatically created and linked all of the Nodes on the terrain that I needed to to actually use a*. I've reworked the whole thing several times now; because everytime I get it all done n' working, I see ways of making it simpler and a bit more efficent. So I really can't express how helpful and informative it was to just get my hands dirty, trying to get the path-finding working! : D



But as I was reading through that AI book, and started dreaming about all of the game-play possibilites that my newfound knowledge had opened up, the more I wanted to create an actual game that could utilize all this totally awesome AI related stuff! For example: A hide-n-go-seek style game played against a computer player. The level could contain small areas that were "hiding-spots". When it was the computer's turn to hide, they could select a hiding spot, find their way to it, and keep track of how good of a hiding-spot that was based on how long it took the other player to find them. Alternatley, when the computer player was 'it', he would look through all of the hiding-spots that he's discovered so far, and when he finds a player that was hiding, makes a mental "note" of it, so that next time he's 'it', he'll be more likely to check that spot before others. Reinforced learning kind of stuff. I realize that there are a million wholes in that, but I'm excited about the over-all idea.

I was also thinking of a game having to do with ants; perhaps something in the style of a Pikmin/RTS game? I've alwasy been interested in programming, and I think that jumping into AI will help me keep myself interested in programming, more precisely: C++. I took an extremely unhelpful C++ course in high-school, way-back when, and I haven't had a real opportunity to do anything with it since, so I can't say enough good stuff about Torque! It has been the perfect arena for my C++ experiments and I still can't believe how cool it all is! :P

Ackk! Enough about me; I'd be very interested to hear what anybody else out there thinks about AI + Torque or if anyone else has/is done something along those lines before? All comments are welcome!