scott wrote:

> Perhaps not, but why not just start with a basic map and see what you 
> can come up with code-wise?  You can probably use Haskell for this and 
> just use text output if you can't do graphics.  Start with a simple 2D 
> map with obstacles and a "target", then you can gradually make it more 
> complex and see how you get on.  Could be quite a fun little project 
> with loads of possiblities for expansion.

Random idea: You know how some software uses an "evolution" algorithm 
where it randomly generates new candidates "similar to" the existing 
one, and gets you to "artificially select" which ones "survive"?

Well, suppose I program a whole bunch of bots, each with an individual 
neural network to control it, and them them "survive" by... not getting 
shot by its competetors? :-D

[Yes, I know, it's completely infeasible in reality. But it's a neat 
idea...]

Post a reply to this message