Research On Concurrent Plans As Programs In State Calculus

AI Planning is the core technology of Artificial Intelligence and one of the most challeng in research domain. AI Planning had achieved rapid development over the past decade, and it was widely used in aerospace, robot control, logistics scheduling, game character design and system modeling. State Calculus is a formalized system for Reasoning about Actions. State Calculus derives from the improvement of Situation Calculus and Fluent Calculus which provides a natural and effective solution for Frame Problem and effective Progression reasoning mechanism for Reasoning about Actions, and has a much broader scope of application. The study of Concurrent action and Plans as Programs are based on State Calculus, and the main works are as follows:(1)The Concurrent Plans as Programs in State Calculus is proposed. The relations of actions of a plan to allow plans that are sequential, non-determined, concurrent, and so on, in which especially a concurrent relation containing interrelated, non-interrelated and interleaving concurrency is provided by introducing Planning Expression. The syntax and semantics for the Planning Expression are presented while overcoming the inherent defects of only interleaving concurrency or synchronized action. A dynamic planning operator of the Concurrent Plans as Programs in State Calculus is proposed. It implements replanning mechanism according to differences of the on-line execution state and previous corresponding off-line execution state and improves dynamic planning operator based on Situation Calculus.(2)The Concurrent Plans as Programs in State Calculus Planner CPAPP-SC is designed and implemented. The composition of CPAPP-SC is given. CPAPP-SC interpreter is implemented by Prolog language according to the semantics of Planning Expression and Search.(3)An improved dynamic planning operator called ESearch is proposed and implemented. The semantics for the Planning Expression and ESearch are presented again, and implemented. ESearch solves the unsuccessful defect of Search on non-determined off-line.