Ignoring path finding and using collaborative diffusion

[icefox]

For my solution I took a look at collaborative diffusion rather than doing any shortest path finding. I ended up writing up a whole fun blog about this strategy (sorry no code or advanced diffusion strategies until after the end of the challenge). So if your tried of implementing A*, shortest path caching, or just looking to check out a different approach:

[ChrisH]

Based on your blog post this sounds like an enhanced version of breadth-first-search, would you agree?

[Narsis]

this strategy is actually the very first thing that came to my mind, but i didn't know much about it so put it off. after reading the blog post(and linked article) i've decided to try this again. one question though:

You mention that the runtime is O(rows*cols). how is that faster then the time required for pathing algorithms?

[icefox]

[erdman]

Interesting. Got link to games played by collaborative-diffusing anti-objecting antbot?

[icefox]

[icefox]

[dabino]

Interestingly, I have started with almost the same approach, though I called it Field / Potential. In my model each object(s) "casts" a field on the mao, which potential is a function of pathlength from the object(s). Then ants just go into direction of potential decline. Now I have up to 7 diferent fields, which define such behaviors like food gathering, attack, defence, exploration etc, and I can combine them accoridng to strategic situation.
Unfortunately, this approach is nothing without good tactical algorythm.

[CrazyMLC]

Man, gotta love emergent solutions.
This means I can be my usual non-program savvy self and still make smart AI, as long as I model the problem right. The ants do all the work!

Only catch: How efficient is this, as far as turn time is concerned?

[pedrosorio]