Planning by Guided Hill-Climbing
This paper describes a novel approach will be called guided hill climbing to improve the efficiency of hill climbing in the planning domains. Unlike simple hill climbing, which evaluates the successor states without any particular order, guided hill climbing evaluates states according to an order recommended by an auxiliary guiding heuristic function. Guiding heuristic function is a self-adaptive and cost effective function based on the main heuristic function of hill climbing. To improve the performance of the method in various domains, we defined several heuristic functions and created a mechanism to choose appropriate functions for each particular domain. We applied the guiding method to the enforced hill climbing, which has been used by the Fast Forward planning system (FF). The results show a significant improvement in the efficiency of FF in a number of domains.
KeywordsSuccessor State Hill Climbing Heuristic Function Heuristic Selection Logistics Domain
Unable to display preview. Download preview PDF.
- 1.Ghallab, M., Nau, D., Traverso, P.: Automated Planning Theory and Practice. Morgan Kaufmann, San Francisco (2004)Google Scholar
- 3.Rich, E., Knight, K. (eds.): Artifice Intelligence. McGraw-Hill, New York (1991)Google Scholar
- 4.Russell, S.J., Norvig, P.: Artifice Intelligence: A Modern Approach. PrenticeHall, Englewood Cliffs (1995)Google Scholar
- 5.Yuret, D., Maza, M.: Dynamic Hillclimbing: Overcoming the Limitations of Optimization Techniques. In: The Second Turkish Symposium on Artificial Intelligence and Neural Networks, pp. 208–212 (1993)Google Scholar
- 6.Juels, A., Watenberg, M.: Stochastic Hill-Climbing as a Baseline Method for Evaluating Genetic Algorithms. Tech-Rep, University of California at Berkeley (1994)Google Scholar
- 7.Rudlof, S., Koppen, M.: Stochastic hill climbing with learning by vectors of normal distributions. Nagoya, Japan (1996), citeseer.ist.psu.edu/rudlof97stochastic.html
- 8.David, P., Kuipers, B.: Learning hill-climbing functions as a strategy for generating behaviors in mobile robots. TR AI90-137, University of Texas at Austin (1990)Google Scholar
- 11.Hoffmann, J.: The Metric-FF planning system: Translating ignoring delete lists to numeric state variables 20, 291–341 (2003)Google Scholar
- 12.Brafman, R., Hoffmann, J.: Conformant planning via heuristic forward search: A new approach. In: Proceedings of the 14th International Conference on Automated Planning and Scheduling (ICAPS-2004), Whistler, Canada (2004)Google Scholar
- 13.Hoffmann, J., Brafman, R.: Contingent planning via heuristic forward search with implicit belief states. In: Proceedings of the 15th International Conference on Automated Planning and Scheduling (ICAPS-2005), Monterey, CA, USA, pp. 71–80 (2005)Google Scholar
- 14.Botea, A., Enzenberger, M., Mueller, M., Schaeffer, J.: Macro-FF: Improving AI Planning with Automatically Learned Macro-Operators 24, 581–621 (2005)Google Scholar