Multiple Behavior Trees for different types of bots?

Hi, I’ve working on a template which consists of 3 types of enemy bots, each with totally different behaviors for threat detection and attacking. At the moment, I’m using the Pawn Sensing component in each of their characters blueprints. However I want to try it out in Behavior Trees as well. But my concern is whether I should have different behavior trees for different types of bots or have a single behavior tree with condition based code. I will be adding atleast 4 other types of bots as well in the near future. So is it advisable to use multiple behavior trees as far as performance is concerned? Will it make much difference if I use just one complex behavior tree?

If the behaviour is actually different in terms of what they do I would use multiple trees. Using different behaviour trees should be much easier to manage and debug. The more behaviour types you add to one tree the more complex one tree will become until you rip your hair out and split them up anyway.

You can replicate a lot of functionality though tasks and services anyway as well as just copy and paste from one tree to another.

If the behaviour differs only in terms of parameters, sight distance, threat weighting etc then it is probably easier to just use one.

And a bunch of simpler BTs will always be better CPU-performance-wise than one big & complex BT. Also, having many BT assets is no problem at all in terms of memory, so I’d recommend going that route.

Quick piggy back related question: Do you ever cache the up front allocation of new instances of different behavior tree assets to avoid any startup costs? I’ve noticed some slight hitches/slowdowns when I have 5 or so bots switch states to a a more complex/higher memoery cost behavior over a one to a few frames.

Thanks a lot guys.