Hi!
I have a question about action atomicity in terms of Utility AI.
Let's say my unit has a melee attack and a default movement ability.
When I evaluate different actions that my AI agent can perform, should I make two separate decisions — “move to X where you probably can hit someone” and “attack X who is right next to you”, or one decision — “move to X and attack Y”?
From my perspective as a player, I don't decide to move somewhere just because there is somebody to hit, and then decide to hit. I decide to move and hit a specific target, non-separably. So, should I keep them as one decision, or separate them but make the "move" action also consider all of the important things about the targets at this point, like health, armor, statuses, etc?
