摘要
基于模型检测的智能规划是当今通用的智能规划研究的热点,其求解效率比较高.但是,目前基于模型检测的智能规划系统没有考虑到利用领域知识来提高描述能力和求解效率.为此,研究了增加领域约束的基于模型检测的智能规划方法,并据此建立了基于模型检测的领域约束规划系统DCIPS(domain constraints integrated planning system).它主要考虑了领域知识在规划中的应用,将领域知识表示为领域约束添加到规划系统中.根据规划=动作+状态,DCIPS将领域约束分为3种,即对象约束、过程约束和时序约束,采用对象约束来表达状态中对象之间的关系,采用过程约束来表达动作之间的关系,采用时序约束表达动作与状态中对象之间的关系.通过在2002年智能规划大赛AIPS 2002上关于交通运输领域的3个例子的测试,实验结果表明,利用领域约束的DCIPS可以方便地增加领域知识,更加实用化,其效率也有了相应的提高.
The MIPS (model checking integrated planning system) has shown distinguished performance in the second and the third international planning competitions. In this paper, a declarative approach to adding domain knowledge in MIPS is presented. And DCIPS (domain constraints integrated planning system) has been developed according to this method. DCIPS allows different types of domain control knowledge such as objective, procedural or temporal knowledge to be represented and exploited in parallel, thus combining the ideas of planning = actions + states into domain control knowledge. An advantage of this approach is that the domain control knowledge can be modularly formalized and added to the planning problem as desired. DCIPS is experimentally verified on the three examples in the transportation domain from AIPS 2002 planning competition where it leads to significant speed-ups.
出处
《软件学报》
EI
CSCD
北大核心
2004年第11期1629-1640,共12页
Journal of Software
基金
国家自然科学基金
国家教育部博士点基金~~
