An Automated Cyclic Planning Framework Based on Plan-Do-Check-Act for Web of Things Composition

Abstract

Empowering end users to be directly involved in the development and composition of their smart devices surrounding them that achieves their goals is a major challenge for End User Development (EUD) in the context of Web of Things (WoT). This can be achieved through Artificial Intelligence (AI) planning. Planning is intended as the ability of a WoT system to construct a sequence of actions, that when executed by the smart devices, achieves an effect on the environment in response to an end user issued goal. The problem of planning specifically for the WoT domain has not been sufficiently dealt with in the existing literature. The existing planning approaches do not deal with one or more of the following important factors in the context of WoT: (1) random unexpected events (2) unpredictable device effects leading to side effects at runtime, and (3) durative effects. In this work, we propose a cyclic planning system which adopted a PDCA (Plan-Do-Check-Act) process solution to deal with the existing shortcomings for continuous improvement. The planner employs domain knowledge based on the WoTDL (Web of Things Description Language) ontology.The cyclic planner enables continuous plan monitoring to cope with inconsistencies with user issued goals. We demonstrate the feasibility of the proposed approach on our smart home testbed. The proposed planner further enhances the ease of use for end users in the context of our goal-oriented approach GrOWTH.