GTD-POP: Bridging the Gap between Soundness and Efficiency in Practical Planners
Daniel M. Gaines · 1996
This paper presents GTD-POP, a planning methodology based on the Generate, Test and Debug paradigm. GTD-POP's goal is to achieve a compromise between planning efficiency and soundness by using associational knowledge to guide its search for interaction checks in a partially-ordered plan. This paper describes the GTD-POP and discusses issues involved in evaluating the behavior of practical planners. Introduction As AI planners leave the realm of toy domains and are brought to bear on complex planning problems, limitations of existing formalisms have become apparent. Designing a domain-independent, sound planner which can work efficiently in complex domains is an infeasible goal. Previous planners have given priority to a few of these characteristics at the expense of others. This paper proposes a model of planning which provides a compromise between soundness and efficiency, allowing a tradeoff between these properties to meet the needs of a given problem. The proposed framework, GTD-P...