So I am working in Unity (5.x) Pro and have my AI agents moving about via calls to Unity's built in path finding and movement system (NavMesh base).
I also, in the past did alot of work with Reynold style steering behaviors, and had alot of nice (uselfull) emergent behaviors come about. But when I used them there was a low density of static obstacles, so a true global path finding solution was unnessisary.
But for this new Unity project, my maps have a high density of obstacles (also some areas of low density, e.g. large courtyards, etc), and a global path finder is quite nessisary (some places can be maze like).
So what I want to do is somehow properly mix NavMesh based path planning with Steering Behaviors (Evade, Chase, Formations, etc.). It is not quite clear to me how that would work.
The NavMesh does give me all walk-able areas, which I can query with ray casts. I also can compute paths from any location on the navmesh to any other location on the navmesh, without having to traverse them (Ive used that to know if even a path exists), but this is an expensive operation and can not be perform too often (like once an update frame, would not work). The number of AI agents I am dealing with right now is anywhere from 1 to 32.
I was just wondering if others have tried to mix these two movement systems, and how they went about it?
Thanks,
-Ryan