AIChallenge - Desirability Field Gradient

I just started on the Fall 2011 Google AI challenge Ants. So far I've implemented a desirability field. It works by having game objects apply gradients to a 2D array of the map. Depending on the object, a negative (food) or positive (water, friendly ants) field is applied. Orders are issued to the ants so that they move towards more negative fields.

Category:

Science & Technology

Tags:

• ai
• ants
• Video Game
• artificial intelligence

[gradbot]

That's the trick. Each point of the map is averaged with its neighbors every turn. This averaging causes field fluctuations to propagate across the map. Each object adds to the field in such a way that the averaging pushes it along and away from where it has been or towards a goal.

[vladimirfol1]

great idea!

i would like to ask: how do you realize exploring of the map?

do u calculate the gradient of all unexplored cells or only of nearest unexplored cells? i think that on big maps allpoints gradient calculating can take a lot of time