Partial and Full Goal Satisfaction in the MUSA Middleware

Classical goal-based reasoning frameworks for agents suppose goals are either achieved fully or not achieved at all: unless achieved completely, the agents have failed to address them. This behavior is different from how people do and therefore is far from real-world scenarios: in every moment a goal has reached a certain level of satisfaction. This work proposes to extend the classical boolean definition of goal achievement by adopting a novel approach, the Distance to Goal Satisfaction, a metric to measure the distance to the full satisfaction of a logic formula. In this paper we defined and implemented this metric; subsequently, we extended MUSA, a self-adaptive middleware used to engineer a heterogeneous range of applications. This extension allows solving real situations in which the full achievement represented a limitation.