Extended Abstract: Recognizing Plan/Goal Abandonment

Introduction This paper describes ongoing efforts to extend our work (Geib & Goldman 2001b; 2001a) (G&G) on the Probabilistic Hostile Agent Task Tracker (PHATT) to handle the problem of goal abandonment. As such, we will be discussing both probabilistic intent inference and the PHATT system and assume the reader is familiar with these areas. We refer the interested reader to the complete paper and to our earlier papers for more discussion of these issues.1 We are applying plan recognition methods, specifically PHATT, to two real-life applications: building assistant systems for elderly patients in assisted living environments and tracking hostile agents in computer network security. Both of these application areas require us to be able to recognize, and reason about, situations when the agents we observe have abandoned a plan or goal. Consider the case of an elder2 who begins the process of taking her medication but gets distracted and does not complete the plan. If a system can recognize when that plan has been abandoned and will not be completed, the system can provide helpful reminders. In general, the ability to infer abandoned goals is an operational requirement for any plan recognition system that is executing incrementally and continuously. Abandoning goals is something that any real observed agent will do. If a plan recognition system is unable to recognize this fact, the system will build up an ever increasing set of active or open plans that the agent has no intention of completing. Without some method of identifying abandoned plans a system attempting to find completions for these open plans will wind up considering unreasonable situations such as the elder begins making a sandwich which they didn’t finish until several hours, days or even weeks later. Existing plan recognition systems do not draw such inferences. Instead a number of different domain assumptions allow them to work around this problem: