不确定层次任务网络规划研究综述

层次任务网络(Hierarchical task network, HTN)规划作为一项重要的智能规划技术被广泛应用于实际规划问题中, 传统的HTN规划无法处理不确定规划问题.然而, 现实世界不可避免地存在无法确定或无法预测的信息, 这使许多学者开始关注不确定规划问题, 不确定HTN规划研究也成为HTN规划研究的前沿.本文从HTN规划过程出发分析了不确定HTN规划问题中涉及的三类不确定, 即状态不确定、动作效果不确定和任务分解不确定; 总结了系统状态、动作效果和任务分解等不确定需要扩展确定性HTN规划模型的工作, 以此对现有不确定HTN规划的研究工作加以梳理和归类; 最后,对不确定HTN规划研究中仍需要解决的问题和未来的研究方向作了进一步展望.

Review on Hierarchical Task Network Planning under Uncertainty

As an important automated planning technique, hierarchical task network (HTN) planning has been widely used in practical planning problems, but traditional HTN planning cannot deal with a planning problem with uncertainty. However, there inevitably exist uncertain or unpredictable information in the real world. As a result, many scholars began to focus on planning under uncertainty, and HTN planning under uncertainty has become the forefront research of HTN planning. In this paper, uncertainties of state, action effects and task decomposition are analysed, followed by a summary of the expansion of HTN planning model for treating these three types of uncertainties. Also, existing research works about HTN planning under uncertainty are reviewed and categorized, based on which some unsolved problems in HTN planning with uncertainty and future research directions are brought forward.