Mining useful macro-actions in planning

Abstract

Planning has achieved significant progress in recent years. Among the various approaches to scale up plan synthesis, the use of macro-actions has been widely explored. As a first stage towards the development of a solution to learn on-line macro-actions, we propose an algorithm to identify useful macroactions based on data mining techniques. The integration in the planning search of these learned macro-actions shows significant improvements over six classical planning benchmarks.

Capability Construction of C4ISR Based on AI Planning

“…Since the heuristic search planner needs to calculate the heuristic value for the candidate action in every iteration, the search efficiency of the planner is greatly decreased when the number of actions increase. Therefore, this study draws on the practice in [21] and [22] to analyze the existing planning solutions, extract the sequence of operations that often appear, and combine them into macro-operations to reduce planning time. The planner structure diagram proposed in this study is shown in FIGURE 6.…”

“…Literally, the addeff(deleff) means that it will be added to (deleted form) the state set after calling the corresponding service. At the same time, Algorithm 2 can be divided into two parts, one is to merge the actions (line 2-21), and the other is to complete the generalization (line [22][23][24]. In the merge part, the precondition and effect of macro operation O i are adjusted when merging the actions.…”