In this paper we discuss intention recognition in general, and the use of logic-based formalisms, and deduction and abduction in particular. We consider the relationship between causal theories used for planning and the knowledge representation and reasoning used for intention recognition. We look at the challenges and the issues, and we explore eight case studies. 1

We are investigating how sensors can improve a portable reminder system (PEAT) that helps individuals accomplish their daily routines. PEAT is designed for individuals who have difficulty remembering when to perform activities because of cognitive impairments from strokes or other brain injuries. These impairments can cause a significantly reduced quality of life for afflicted individuals and their caregivers. PEAT provides assistance by planning a schedule of activities for a user, and by cueing the user when activities should begin or end. One limitation of PEAT is that it requires the user to manually indicate when an activity starts or stops, which causes unnecessary cues for a user who needs only occasional reminders. By incorporating feedback from reliable sensors, the software can automatically infer which activity the user is performing. With this information, we expect that PEAT will be able to cue the user more effectively, by not cueing the user when sensors indicate that activities have already started or stopped, and by providing compliance cues that remind the user when steps of an activity have been forgotten. We present a description of the system, implemented scenarios, and a discussion of potential benefits and pitfalls with this approach.

using temporal features to create more