Creating AI Planning Domains for Smart Environments Using PDDL

One corner stone for user assistance in smart environments is service composition, which can be reached using AI planning. Therefore it is required to formulate the current environment as planning problem, which can than lie solved with an AT planning algorithm. However, sonic characteristics of smart environments conflict with the strict requirements of classical AI planning as well as with the limitations of PDDL, which is the state of the art representation language for planning domains and problems. These interferences are uncovered in a. bottom up manner by reviewing smart environment scenarios from literature. The scenarios were realized as planning problems ill PDDL, consisting of operators, an initial world state, propositions, and goals. Possible problems and ways to avoid the respective hurdles are given. It is concluded if AI planning based user assistance is suitable to solve the crucial challenges of the respective scenario and to what extend AI planning is able to contribute to the problem of service composition in smart environments in general.