Complete contingency planners
Dennis F. Kibler, Karl Schwamb · eScholarship (California Digital Library) · 1992
A framework is proposed for the investigation of planning systems that must deal with bounded uncertainty. A definition of this new class of contingency planners is given. A general, complete contingency planning algorithm is described. The algorithm is suitable to many incomplete information games as well as planning situations where the initial state is only partially known. A rich domain is identified for the application and evaluation of contingency planners. Preliminary results from applying our complete contingency planner to a portion of this domain are encouraging and match expert level performance. 1. Introduction Traditional AI planning approaches utilize simplified models of the world in which all information needed to plan is known and all actions are performed solely by the planning agent. While these models stimulated research for a number of years, it is increasingly clear that they ignore important issues critical to real-world applications. Dissatisfaction with this...