...
Arguably the most central element of a flipper autopilot is the module that decides when to trigger the flippers. In this system said descision decision is mainly made in the region displayed above. Most of the time while playing the automaton will be in the survive state, where it reacts to the times as given by the prediction. Here the only objective is to keep the ball in the game.
In certain situtations situations this state is left in order to perform more elaborate gameplay maneuvers.
One of these is to catch the ball with the left flipper. Once this was successfully performed targeted shots to certain game elements are possible and the controller decides which one to aim for by analyzing some lamps that indicate certain game situations. The rule of thumb is: When there is a lamp lit above some target, it should be aimed at.
Another one is to detect a ball coming back after successfully hitting the big ramp. The ball is then led down a deterministic path feeding it into the left inlane. In this situation the controller tries to trigger the left flipper at the right timing to shoot the ball right into the big ramp again. This is useful since successive ramp shots are benefitial beneficial in most situations, as they score good points and lead to activation of a multiball mode.
...
This region simply triggers the top flipper whenever there is a ball detected to be close to its range. Hitting the ball with the top flipper is often benefitial beneficial but never critical. So in order to save resources there is less computational effort put into it, i.e. no prediction is performed and the range is not precisely defined. It is implemented as a single state with a during action, since the work of detecting, whether there is a ball near the top flipper, is performed by the image processing, which sets a flag in the respective case.
...