@aron_granberg I have read a lot of different posts for flying AI and I understand that it’s out of the scope of this because of performance limitations but I have an interest on how it should be properly solved with some ideas:
I was thinking that a very simple and seemingly optimized solution would be to create a cube as you have now for the area of effect and create points in there where each knows it’s neighbor, then use the A* to traverse through those nodes an use em to fly there.
Having a parent object to change the Y is an ok solution but it creates various issues along the way and it’s not the best architecture if that makes sense, I tried something similar with changing the Y on LateUpdate() which worked nicely but there were issues as in I had to create a different graph for the overground areas which isnt a problem but if I had say a tall bridge and the graph wasnt on that since they would fly, the enemy not finding the spot I was on would just stay where he was and that’d be it.
In any case I was mostly wondering what do you think of the 3D neighboring points solution?