Ahh! thanks for the info - I didn't know much of that. Oh well it's not off topic really - many great advancements in robotics and AI have come from mimicry of the animal kingdom. The 'flow field' biasing I described seems to my mind to be an analogue of ants leaving scents - perhaps there's something useful in the bee technique?
For example instead of storing entire paths between points - you could just store the distance travelled by an entity between those two points - then when pathing to that point - compare the linear distance to the distance travelled and if it's greater than you know that there's obstacles between the start and end points. I don't know how that can be useful exactly - but perhaps store the distance travelled and a series of left/right indicators - then when the entity encounters an obstacle it pops a left/right off the stack and knows which way to turn.
To my mind it would result in an effective path followed between two points across a static terrain but with far less memory used to store that path.
it would also plug nicely into a steering algorithm. Just a thought experiment but an interesting one!
Bookmarks