9619579 Pollack This is the first-year of a three continuing award. Research on planning within AI has led to the development of computational techniques for generating a plan to achieve a specified goal from a specified initial state, given definitions of the available actions. Until recently, however, much of this research has been governed by a number of simplifying assumptions, notably that there is no indeterminacy either in the planning agent's knowledge, in the effects of its actions, or in the notion of goal satisfaction, and that there are no changes either in the goals presented to the agent or in its environment. This award supports research aimed at relaxing the latter assumption. Specifically, the research investigates the issues involved in extending planning technology to a dynamic setting, in which the agent's goals and environment are both subject to change. A study is being made of both the theoretical and practical issues presented by situations that call for integrated generation, maintenance, and execution of plans for multiple, changing goals. The project addresses these issues from two complementary points of view. The first is designing and implementing a dynamic planning system that is flexible enough to allow for the experimental investigation of the consequences of adopting different reasoning strategies in different dynamic planning environments. Secondly, but in parallel, a formal analysis is being carried out of the dynamic planning problem, with a special emphasis on providing a clean semantic analysis of the implemented system. The aim of the project is more flexible and realistic planning models and systems.
