Hi,
I’m using a recast graph in 2D with AIPath. As you can see in the picture, I would like the red agent (AI) to avoid the blue object (e.g. player) “as much as possible” but not like it’s an impassable barrier - if forced in a bad position, it should go through the circle, but still as far from the blue object as possible.
What is the best way to achieve this? From what I can tell area penalties are fixed and static. Local avoidance seems to only consider other agents? Or should I write a custom heuristic / penalty system somewhere?
Btw, is the distance based penalty even possible with navmeshes? If I had to penalize a node it would be one value for the entire node, which wouldn’t make sense since they’re all differently sized.