Plan recognition in smart environments

Abstract

This position paper describes our approach to recognizing tasks of mobile users in a smart environment. Such high-level interpretation of behavior enables context-aware applications to adapt to the users' needs and intentions. In the AI community, plan recognition techniques have proven their applicability in recognizing the tasks of software agents in a controlled environment. However, these approaches fall short in recognizing tasks of people in a real-world environment. Therefore, we propose several extensions to plan recognition techniques by using constraints and hybrid reasoning algorithms. In addition, we propose to improve the plan recognition process with multi-step processing of context information. We also discuss how our approach leverages some of the difficulties of plan recognition in smart environments.

Action Prediction in Smart Home Based on Reinforcement Learning

“…All the new incoming events are recorded in new plans and saved in plan library. In [15,16] they define plan library as a set of plans or predefined steps. In this proposition plan library is constructed online through monitoring the user.…”