I have several Smart NavLinks that lets AI to jump over obstacles.
On PIE start AI plans to go around an obstacle (I visualize it with FindPathToLocationSynchronously node)
But when pawn gets close to it, all of sudden it makes very late decision to jump over it. Even if it means to turn 90 degrees to side for that.
My nav mesh is static as is my environment. It still happens with only one pawn, so influence of other pawns can’t be the reason.
So question is, is it possible to let AI consider NavLink paths earlier (some of them do however, so make it more precise?), or make them stick to their initial path until they reach the target? (There is nothing that needs them to reconsider their paths midway)
Ask me if there anything that I forgot to clarify.
