ABSTRACT
The challenge we address is to reason about projected resource usage within a hierarchical task execution framework in order to improve agent effectiveness. Specifically, we seek to define and maintain maximally informative guaranteed bounds on projected resource requirements, in order to enable an agent to take full advantage of available resources while avoiding problems of resource conflict. Our approach is grounded in well-understood techniques for resource projection over possible paths through the plan space of an agent, but introduces three technical innovations. The first is the use of multi-fidelity models of projected resource requirements that provide increasingly more accurate projections as additional information becomes available. The second is execution-time refinement of initial bounds through pruning possible execution paths and variable domains based on the current world and execution state. The third is exploitation of additional semantic information about tasks that enables improved bounds on resource consumption. In contrast to earlier work in this area, we consider an expressive procedure language that includes complex control constructs and parameterized tasks. The approach has been implemented in the SPARK agent system and is being used to improve the performance of an operational intelligent assistant application.
- K. R. Apt. The essence of constraint propagation. Theoretical Computer Science, 221(1--2):179--210, 1999.]] Google ScholarDigital Library
- B. J. Clement, A. C. Barrett, G. R. Rabideau, and E. H. Durfee. Using abstraction in planning and scheduling. In Proc. of ECP'01, 2001.]]Google Scholar
- M. Dastani and L. van der Torre. Specifying the merging of desires into goals in the context of beliefs. In Proc. of EurAsia ICT '02, 2002.]] Google ScholarDigital Library
- L. P. de Silva and L. Padgham. Planning on demand in BDI systems. In Proc. of ICAPS '05 Poster Session, pages 37--40, June 2005.]]Google Scholar
- R. Dechter. Constraint Processing. Morgan Kaufinann, San Francisco, CA, May 2003.]] Google ScholarDigital Library
- B. Drabble and A. Tate. The use of optimistic and pessimistic resource profi les to inform search in an activity based planner. In Proc. of AIPS '94, pages 243--248, Chicago, IL, June 1994.]]Google Scholar
- M. P. Georgeff and A. L. Lansky. Procedural knowledge. Proc. of the IEEE, 74(10):1383--1398, 1986.]]Google ScholarCross Ref
- P. Laborie. Algorithms for propagating resource constraints in AI planning and scheduling. Artificial Intelligence, 143(2), 2003.]] Google ScholarDigital Library
- D. Morley and K. Myers. The SPARK agent framework. In Proc. of AAMAS '04, pages 714--721, New York, NY, July 2004.]] Google ScholarDigital Library
- N. Muscettola. Computing the envelope for stepwise-constant resource allocations. In Proc. of CP '02, pages 139--154, Sept. 2002.]] Google ScholarDigital Library
- K. L. Myers and N. Yorke-Smith. A cognitive framework for delegation to an assistive user agent. In AAAI 2005 Fall Symposium on Mixed-Initiative Problem-Solving Assistants, Nov. 2005.]]Google Scholar
- A. Pfeffer. Functional specifi cation of probabilistic process models. In Proc. of AAAI'05, pages 663--669, Pittsburgh, PA, July 2005.]] Google ScholarDigital Library
- A. S. Rao and M. P. Georgeff. Modeling agents within a BDI-architecture. In Proc of KR'91, pages 473--484, 1991.]]Google Scholar
- SRI International. CALO: Cognitive Assistant that Learn and Organizes. www.ai.sri.com/project/CALO, Mar. 2005.]]Google Scholar
- J. Thangarajah, L. Padgham, and J. Harland. Representing and reasoning for goals in BDI agents. In Proc. of the Australasian Conference on Computer Science, Melbourne, Australia, Jan. 2002.]] Google ScholarDigital Library
- J. Thangarajah, M. Winikoff, L. Padgham, and K. Fischer. Avoiding resource conficts in intelligent agents. In Proc. of ECAI-02, 2002.]]Google Scholar
Index Terms
- Continuous refinement of agent resource estimates
Recommendations
Resource Load Balancing Based on Multi-agent in ServiceBSP Model
ICCS '07: Proceedings of the 7th international conference on Computational Science, Part III: ICCS 2007Based on ServiceBSP model, a resource load balancing algorithm with Multi-Agent is put forward in this paper which achieves the goal of dynamic load balancing and favorable fault-tolerant. The algorithm calculates the load value according to the status ...
Design of Multi Agent System for Resource Allocation and Monitoring
The objective of Resource Allocation and Monitoring System is to make the procedures involved in allocating fund resources to competing clients transparent so that deserving candidates get funds. Proactive and goal directed behaviour of agents make the ...
Comments