For the game I'm working on, the player has the ability to throw certain items onto the battle board that can effect the AI. By way of example, one of the items is a decoy. If the player throws a decoy on the board, then for the number of actions the item is active (10 for a default decoy), the AI will move toward the decoy instead of the player.
Some AI will be effected by some items, and not by others. I'm looking for a nice clean way of handling getting the information of what Items are currently on the board to each AI to allow them handle the specific items that effect them. I'm thinking some sort of consumer pattern, where each enemy type can "subscribe" to the item types that effect them and get notified when ever any of those are on the board.
If anyone could point me to information on any kind of pattern that would handle this situation, that would be most appreciated. I'm writing the game in Java, but examples in any language are fine.
I suggest turning your idea of collecting influences around by calling, on appropriate AI agents, suitable methods to change their behaviour (decoy example: "track object with a certain ID (i.e. the decoy) until further notice") whenever each item appears, disappears, loses potency, etc. Indexing AI agents by type, location etc. might be useful, but the items can be responsible for filtering the set of affected AI agents.
Work should be minimized because processing would happen once per item-related event when "influences" change, i.e. once in a few turns, rather than once per turn when AI needs to manage its own state, and possibly on a reduced subset of AI agents rather than all.
Moreover, there are reasons of principle: the items you describe should depend on AI (the game can work without items), but the AI layer shouldn't be aware of such a superstructure; secondarily, methods allowing rather authoritative outside control over AI behaviour are a reasonable feature that doesn't threaten the integrity of the AI and can be useful for level scripting and other purposes, while burdening AI agents with tracking influences is brittle and complex.
Don't worry too much about what items affect what AI agents; you can conservatively give orders to agents the item isn't sure about.
For example, if fixed things like traps "understand" (as they should) that they aren't going to chase anything around the map, if they receive a decoy's request to be attacked they won't be confused.
That sounds like a good direction to go in. Items can then have a nice area of influence, and pick any AI within that area and apply their effect to those objects. Then as you said, the AI object would know what effects it responds to or what one it ignores.