·24· 智能系统学报 第2卷 统在对象、关系、状态、过程、行为等方面存在相似 [8]TA TE A.Generating project networks [A].In Proceed- 性，这将成为更全面、更深入的领域描述机制建立的 ings of UCAI-77[C].Boston,Mass,USA,1977 基础 [9]SACERDOTI E.A Structure for Plans and Behavior 3)从元层次研究协同的求解策略 M].New York:Elsevier,1977. [10]EROL K,HENDL ER J,NAU D.Semantics for hier- 多策略的协同求解是应用智能系统的重要求解 archical task-network planning[R].CS TR-3239,UMI 策略.从元层次进行策略的协同应成为一个重要的 ACS TR-94-31,ISR TR-95-9,University of 研究方向.虽然元层次的求解已经在规划研究中得 Maryland,1994. 到应用，但仍然在起步阶段，目前尚无比较系统的研 [11]EROL K,HENDL ER J,NAU D.HTN planning: 究，尤其是利用元知识进行协同求解的研究.基于元 complexity and expressivity [A ]In Proceedings of 知识的求解策略研究将为多求解策略的建立、维护、 AAAF94 [C].Seattle,USA,1994. 扩展提供顶层的方法和指导 [12]NAU D,AU T,IL GHAMI O,et al.SHOP2:an 5结束语 HTN planning system[J ]JAIR,2003,20:379-404. [13 ]SIMPSON R,MCCLUSKEY T,LIU D,et al.Knowl- 综上所述，智能规划是兼具理论和应用价值的 edge representation in planning:a PDDL to OCLh trans- 研究领域，经过多年的发展，其研究与应用已经涉及 lation [A].ISMIS2000 [C].Charlotte,USA,2000. 到众多的领域，形成了多种理论体系并在实践中得 [14]TATE A,DRABBLE B,DAL TON J.O-plan:a 到体现.但在大规模、复杂领域中，目前的理论和方 knowledge-based planner and its application to logistics [A].Advanced Planning Technology[C].Morgan Kauf- 法还远远不够.面向实际应用问题是充分应用现有 mann,1996. 理论、技术并提出新的理论问题的重要途径.应用研 [15]WIL KINS D.Using the SIPE2 planning system:a 究应该结合应用领域的问题特征进行，并将知识在 manual for version 6.0[M].SRI International Artificial 多角度、多层面的运用作为提高系统实用性的关键 Intelligence Center.Menlo Pork.California.1999. 技术进行全面、深入的研究，并应注重多种知识的集 [16]CHIEN S,RABIDEAU G.ASPEN-automated planning 成与协同 and scheduling for space mission operations [A ] SpaceOps2000 [C].Toulouse,France,2000. 参考文献： [17]CESTA A,ODDI A.A representation language for do- [1]WELD D.Recent advances in AI planning[J ]AI Maga- main knowledge in planning architectures [A].In Pro- ine,1999,20(2):93-123. ceedings of KAW-96 [C].Alberta,Canada,1996. [2]WIL KINS D,DESJARDINS M.A call for knowledge- [18]BLUM A,FURST M.Fast planning through planning based planning [J ]AI Magazine,2001,22(1):99. graph analysis [A].In Proceedings of DCAF95 [C]. 115. Quebec,Canada,1995. [3]BL YTHEJ.An overview of planning under uncertainty [19]HOFFMANNJ.FF:the fast-forward planning system [J].A1 Magazine,1999,20(2):37-54. [U].AI Magazine,2001,22(3):57-62. [4]ZIMMERMAN T,KAMB HAMPA TI S.Learningas- [20]KAUTZ H,SEL MAN B.Pushing the envelope:plan sisted automated planning [J ]AI Magazine,2003,24 ning,propositional logic,and stochastic search [A ]In (2):73-96. Proceedings of AAAF96 [C].Portland,USA,1996. [5 ]FIKES R,NILSSON N.Strips:a new approach to the [21]BERTOLI P,CIMATTIA,SLANEYJ,et al.Solving application of theorem proving to problem solving [J]. power supply restoration problems with planning via Artificial Intelligence,1971,2(3,4):189-208. symbolic model checking [A].In Proceedings of ECA102 [6]MCDERMOTT D.PDDL-the planning domain definition [C].Lyon,France,2002. language EB/OL ]www.cs.yale.edu./homes/dvm, [22 ]J ENSEN R,VELOSO M.OBDD-based deterministic 1998-05.10. planning using the UMOP planning framework [A ]In [7]FOX M,LONGD.PDDL2.1:an extension to PDDL for Proceedings of AIPS-00 Workshop on Model-Theoretic expressing temporal planning domains [J ]Journal of AI Approaches to Planning [C].Breckenridge,USA, Research,2003,20:61-124 2000. 1994-2009 China Academic Journal Electronic Publishing House.All rights reserved.http://www.cnki.net统在对象、关系、状态、过程、行为等方面存在相似 性 ,这将成为更全面、更深入的领域描述机制建立的 基础. 3) 从元层次研究协同的求解策略 多策略的协同求解是应用智能系统的重要求解 策略. 从元层次进行策略的协同应成为一个重要的 研究方向. 虽然元层次的求解已经在规划研究中得 到应用 ,但仍然在起步阶段 ,目前尚无比较系统的研 究 ,尤其是利用元知识进行协同求解的研究. 基于元 知识的求解策略研究将为多求解策略的建立、维护、 扩展提供顶层的方法和指导. 5 结束语 综上所述 ,智能规划是兼具理论和应用价值的 研究领域 ,经过多年的发展 ,其研究与应用已经涉及 到众多的领域 ,形成了多种理论体系并在实践中得 到体现. 但在大规模、复杂领域中 ,目前的理论和方 法还远远不够. 面向实际应用问题是充分应用现有 理论、技术并提出新的理论问题的重要途径. 应用研 究应该结合应用领域的问题特征进行 ,并将知识在 多角度、多层面的运用作为提高系统实用性的关键 技术进行全面、深入的研究 ,并应注重多种知识的集 成与协同. 参考文献 : [1 ]WELD D. Recent advances in AI planning[J ]. AI Maga2 zine , 1999 ,20 (2) : 93 - 123. [2 ]WIL KINS D , DESJ ARDINS M. A call for knowledge2 based planning [J ]. AI Magazine , 2001 , 22 ( 1) : 99 - 115. [3 ]BL YTHE J. An overview of planning under uncertainty [J ]. AI Magazine , 1999 ,20 (2) : 37 - 54. [4 ] ZIMMERMAN T , KAMB HAMPA TI S. Learning2as2 sisted automated planning [J ]. AI Magazine , 2003 , 24 (2) : 73 - 96. [5 ] FIKES R , NILSSON N. Strips: a new approach to the application of theorem proving to problem solving [J ]. Artificial Intelligence , 1971 , 2 (3 ,4) : 189 - 208. [6 ]MCDERMO TT D. PDDL2the planning domain definition language [ EB/ OL ]. www. cs. yale. edu. / homes/ dvm , 1998 - 05 - 10. [ 7 ]FOX M , LON G D. PDDL2. 1 : an extension to PDDL for expressing temporal planning domains [J ]. Journal of AI Research , 2003 , 20 : 61 - 124. [8 ] TA TE A. Generating project networks [A ]. In Proceed2 ings of IJCAI - 77[C]. Boston , Mass , USA , 1977 [9 ] SACERDO TI E. A Structure for Plans and Behavior [ M ]. New York : Elsevier , 1977. [10 ] EROL K , HENDL ER J , NAU D. Semantics for hier2 archical task2network planning[ R]. CS TR23239 , UMI2 ACS TR - 94 - 31 , ISR - TR - 95 - 9 , University of Maryland , 1994. [11 ] EROL K , HENDL ER J , NAU D. H TN planning : complexity and expressivity [ A ]. In Proceedings of AAAI294 [C]. Seattle , USA , 1994. [ 12 ] NAU D , AU T , IL GHAMI O , et al. SHOP2 : an H TN planning system[J ]. J AIR , 2003 ,20 :379 - 404. [ 13 ]SIMPSON R , MCCLUSKEY T , L IU D , et al. Knowl2 edge representation in planning : a PDDL to OCLh trans2 lation [A ]. ISMIS’2000 [C]. Charlotte , USA , 2000. [14 ] TA TE A , DRABBL E B , DAL TON J. O2plan : a knowledge 2based planner and its application to logistics [ A ]. Advanced Planning Technology[C]. Morgan Kauf2 mann , 1996. [15 ]WIL KINS D. Using the SIPE22 planning system : a manual for version 6. 0 [ M ]. SRI International Artificial Intelligence Center. Menlo Pork. California. 1999. [16 ]CHIEN S , RABIDEAU G. ASPEN2automated planning and scheduling for space mission operations [ A ]. SpaceOps2000 [C]. Toulouse , France , 2000. [17 ]CESTA A , ODDI A. A representation language for do2 main knowledge in planning architectures [ A ]. In Pro2 ceedings of KAW296 [C]. Alberta , Canada , 1996. [18 ]BLUM A , FURST M. Fast planning through planning graph analysis [ A ]. In Proceedings of IJCAI295 [ C ]. Quebec , Canada , 1995. [19 ] HOFFMANNJ. FF : the fast2forward planning system [J ]. AI Magazine , 2001 ,22 (3) : 57 - 62. [20 ] KAU TZ H , SELMAN B. Pushing the envelope : plan2 ning , propositional logic , and stochastic search [ A ]. In Proceedings of AAAI296 [C]. Portland , USA , 1996. [21 ]BERTOL I P , CIMA TTI A , SLAN EYJ , et al. Solving power supply restoration problems with planning via symbolic model checking [ A ]. In Proceedings of ECAI02 [C]. Lyon , France , 2002. [22 ]J ENSEN R , V ELOSO M. OBDD2based deterministic planning using the UMOP planning framework [ A ]. In Proceedings of AIPS200 Workshop on Model2Theoretic Approaches to Planning [ C ]. Breckenridge , USA , 2000. ·24 · 智 能 系 统 学 报 第 2 卷