Research On Heuristics With Precedence Constraints For Temporal Planning

Planning is the problem of finding a sequence of actions to reach the goal from the giveninitial state. In a long time, researchers focused on classic planning in which someassumptions are made for simplifying.1cost for each action and no time span are two ofthose assumptions. However, some new types of planning problem are formed by removingsome assumptions for practical application. The approaches based on heuristic have beenplaying an important role in planning search. In this paper, we introduced the precedenceconstraints to the temporal planning, which admits durative actions and numeric resources,and expanded the temporal context-enhanced additive heuristic to obtain a predominatedheuristic. We also explored the orders between atoms using landmarks, and then appliedthose orders to extract the precedence constraints between conditions of actions. Particularly,for conditions relative to comparison variables, we proposed some rules to mine theirprecedence constraints. Several methods are experimented on the given planning domainsand brief analysis is given out. Through the experimental results, we can see the efficiencyand comparability of our heuristic. The potential foreground of the proposed heuristic is alsoshown from the experiments.