D. E. Wilkins and K. L. Myers. A Multiagent Planning Architecture. In Proc. of AIPS-98, pages 154--162, 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....advantages of heuristically guided search with random search. Our ultimate goal is to automate the scheduling execution cycle of a single autonomous agent that needs the services of other agents to accomplish its task. Pollack s DIPART system [22] and the Multiagent Planning architecture (MPA) [31] assume multiple agents that operate independently but all work toward the achievement of a global goal. Our agents are trying to achieve their own goals and to maximize their profit; there is no global goal. 7 Conclusions and Future Work Auction mechanisms are an e#ective approach to ....

David E. Wilkins and Karen L. Myers. A multiagent planning architecture. In Proc. Int'l Conf. on AI Planning Systems, pages 154--162, 1998.

....for representing procedural knowledge common to both SIPE and the plan execution system to which it is linked, PRS. It is argued that an ACT structure corresponds both to a planning operator or plan fragment in a generative planner, and to an operating procedure in a plan execution system [24], and as this language suggests is some way from a problem oriented modelling language. The ACT s editor supports the AI expert, but cannot be said to support a domain expert. OPLAN s modelling language, TF (task Formalism) was supported only by a user manual incorporating heuristics derived from ....

D. Wilkins and K. Myers. A Multiagent Planning Architecture. In Procceedings of AIPS, pages 154--162, 1998. 15

....provide di erent characteristics useful for a common goal [2] But this is not the only example. For instance, it has been shown that di erent planners work well in di erent domains [39] Therefore in some cases it would be a good idea to combine di erent planner agents in the same mas system [41]. They o er modularity, exibility, and adaptability. A mas uses a common language to communicate heterogeneous agents. Hence it is easy to add new agents with new abilities, if required. These characteristics are essential in complex, large or unpredictable domains [35] mas are inherently ....

....for integrating diverse technologies into a system capable of solving complex planning problems. MPA has been designed for application to planning problems that cannot be solved by individual systems, but rather require the coordinated e orts of a diverse set of technologies and human experts [41, 42]. CMUexpress is a mas architecture developed at CMU whose purpose is to plan, execute plans, and monitor its performance. It has been applied to Non combatant Evacuation Operations (NEO) In this particular case, the whole system integrates about 20 agents. In particular, it includes MMM (a ....

D. E. Wilkins and D. L. Myers. Multiagent planning architecture. In Proceedings on The Fourth International Conference on Articial Intelligence Planning Systems. AIPS98, June 1998.

....Force Research Laboratory Planning Initiative (ARPI) Tate, 1996b) The Planning and Decision Aid (PDA) element is a joint project with SRI International. To complement AIAI s work on MOUT, SRI International is exploring the use of, SIPE (Wilkins 1988) PRS (Georgeff Lansky, 1987) and MPA (Wilkins and Myers 1998) in Army operations down from battalion level and in open terrain with mechanized forces. The remainder of this paper is structured as follows. We first introduce the planning requirements of US Army small units and detail how they vary depending upon the phase of an operation and the tempo of ....

....plugins such as INSPECT (Valente et al. 1999) simulation systems (Cohen, Anderson, and Westbrook, 1997) and critics or plan comparison functions. Such sophisticated evaluators were used with the process panel technology included here for an Air Campaign Planning system under the ARPI program (Wilkins and Myers, 1998). Monitoring Plan Execution After a set of plans have been generated, a plan may be selected for execution. While the actions in a plan are being carried out, a situational awareness system can gather information from sensors, or reports by the units that are participating in the operation, in ....

Wilkins, D., and Myers, K., 1998, A Multi Agent Planning Architecture, Proceedings of the 4th International Conference on AI Planning System, Pittsburgh, USA.

....planning extends and generalizes the failure reasoning of critics to general meta reasoning about several other strategies and introduces the blackboard mechanism for communication. For planning in complex domains that differ considerably from the proof planning domains, Wilkins and Myers [23] describe a multi agent planning architecture (MPA) that integrates a meta reasoning component (meta planning cell) and various stand alone problem solving components for the same reasons explained above. 4] propose a multi agent architecture to combine ATPs and proof planning. They propose to ....

D.E. Wilkins and K.L. Myers. A multiagent planning architecture. In Proc. of the Fourth International Conference on AI Planning Systems (AIPS'98), pages 154--162, 1998.

....and it occurs at run time. Each of these assumptions can be contrasted to assumptions that have been made in prior work in Distributed Artificial Intelligence and Multi Agent Systems (George#, 1983; Durfee Lesser, 1987; Genesereth, Ginsberg, Rosenschien, 1986; Tambe, 1997; Corkill, 1979; Wilkens Myers, 1998). Improvement in the performance of the community occurs over several episodes of community activity. During step 2, individuals are satisfied with any solution. In step 3 the actors store into memory the portions of the behavior that they believed were essential to the solution. In future ....

Wilkens, D. and Myers, K. (1998). A multiagent planning architecture. In The Fourth International Conference on Artificial Intelligence Planning Systems, pages 154--162.

....is now a PRSlite (Myers, 1996) and a number of tool benches and formal methods for proving code correct (e. g Huber, 1999; d Inverno et al. 1997) The original development lab for PRS, SRI, is now focusing effort on a much more modularized AI architecture, built under a multi agent paradigm (Wilkins and Myers, 1998). The pre history of PRS, the Shakey project, also has relevant evolutionary trends (Nilsson, 1984) Although Shakey had a traditional planner (called STRIPS) over the term of the project the concept of triangle tables was developed. A triangle table decomposes a plan into its steps and ....

Wilkins, D. E. and Myers, K. L. (1998). A multiagent planning architecture. In Proceedings of AIPS-98, pages 154--162.

....or control relationships are present among the agents. They have each objectives to solve and they cooperate with one another to achieve their goals. We call this type of cooperation capabilitybased. From this prospective, our enterprise is very different from the planning architecture proposed in [19], where the planning process is distributed among agents, however they are centrally controlled. 9 Conclusion Planning in a dynamic open MAS imposes a combination of problems that range from partial domain information to dynamism of the environment. These problems are each resolved, separately, ....

David E. Wilkins and Karen L. Mayers. A multiagent planning architecture. In Proceedings of the Fourth International Conference on Artificial Intelligence Planning Systems (AIPS-- 98), 1998.

....the focus) to previously dismissed search directions. Finally, our approach provides a declarative agent speci cation framework supporting the de nition, deletion, and modi cation of agent societies at run time. Our work is also in uenced by the multi agent planning framework presented in [ Wilkins and Myers, 1998 ] 5 Conclusion In this paper we presented an approach to agent based reasoning. Our framework is based on concurrent suggestion agents working for natural deduction rules, tactics, methods, and specialised external reasoning systems. The suggestions by the agents are evaluated after they are ....

D. E. Wilkins and K. L. Myers. A Multiagent Planning Architecture. Proc. of the Fourth International Conference on Articial Intelligence Planning Systems, 1998. AAAI Press.

....13 about constraints in order to support the active postponement of decisions. It represents each postponed planning decision by a variable that can be constrained during planning. For planning in very complex domains that differ considerably from the proof planning domains, Wilkins and Myers [WM98] present a multiagent planning architecture (MPA) that integrates a meta reasoning component (meta planning cell) and various stand alone problem solving components. Although MPA is similar to MULTI in several aspects (e.g. it also separates out modules that were part of the Sipe planner ....

D.E. Wilkins and K.L. Myers. A multiagent planning architecture. In R. Simmons, M. Veloso, and S. Smith, editors, Proceedings of the Fourth International Conference on AI Planning Systems (AIPS'98), pages 154--162, 1998. 15

....procedures by MOVERSWORLD actors, is shaped by the fact that communication is the central mechanism for cooperation and coordination. Communication is considered expensive , so it occurs at runtime and is never part of the planning process; this is unlike standard planning systems (Corkill, 1979; Wilkens Myers, 1998). Actors limit themselves to single requests; transmitting single requests lowers both communication costs and plan merging costs, which can be considerable for distributed, independent actors. Two actors are said to cooperate if they act or work together to achieve some common purpose and they ....

Wilkens, D. and Myers, K. (1998). A multiagent planning architecture. In The Fourth International Conference on Artificial Intelligence Planning Systems, pages 154--162.

....search. As a result, a solution can still be found even if the focus of the search is misplaced. Clearly, more resources are necessary in the case of a bad than of a good focus. We currently realise the so called focused proof search as an adaptation of the multi agent planning architecture, MPAWilkins and Myers (1998), in the proof planning domain. Important infrastructure for this enterprise is provided by the Omega MEGA (http: www.ags.uni sb.de omega ) proof development environment. The main component of MPA is a multi agent proof planning cell, which consists of 1) several planning agents, 2) a plan ....

D. E. Wilkins and K. L. Myers. A Multiagent Planning Architecture. Proceedingsof AIPS'98, 1998. AAAI Press, Menlo Park, CA, USA.

....first step towards a distributed planning scheme based on a peer to peer cooperation between agents that is not based on hierarchy or control relationships. We call this type of cooperation capability based. In this respect, our approach is very different from the planning architecture proposed in [18], where the planning process is centralized, but the execution distributed. The resulting collaborative process is consistent with the theoretical framework layed out by Grosz et al. [8] and Cohen et al. [2] which concentrates on the commitment of the agent to the team activity as a means to ....

David E. Wilkins and Karen L. Mayers. A multiagent planning architecture. In Proceedings of AIPS-98, 1998.

....of a specialised problem solver is estimated with respect to the updated global proof tree and according to the information communicated between the subsystems. 4. 1 System Components We realise the so called focused proof search as an adaptation of the multi agent planning architecture, MPA, WM98] in the proof planning domain. The main component of MPA is a multi agent 10 4 ARCHITECTURE proof planning cell, which consists of several planning agents, a plan server, a domain server, and finally a planning cell manager. The comparison between MPA in [WM98] and our architecture is given in ....

....planning architecture, MPA, WM98] in the proof planning domain. The main component of MPA is a multi agent 10 4 ARCHITECTURE proof planning cell, which consists of several planning agents, a plan server, a domain server, and finally a planning cell manager. The comparison between MPA in [WM98] and our architecture is given in 7. 4.1.1 Planning Agents To build up a concrete multi agent proof planning cell we choose a finite subset of agents A : fA 1 ; Ang from a given pool of planning agents. A planning agent either consists of a single reasoning system or of a composition ....

[Article contains additional citation context not shown here]

David E. Wilkins and Karen L. Myers. A Multiagent Planning Architecture. In Reid Simmons, Manuela Veloso, and Stephen Smith, editors, Proceedings of the Fourth International Conference on Artificial Intelligence Planning Systems (AIPS'98), pages 154--162, Pittsburgh, PEN, USA, June 7--10 1998. AAAI Press, Menlo Park, CA, USA.

....of a specialised problem solver is estimated with respect to the updated global proof tree and according to the information communicated between the subsystems. 4. 1 System Components We realise the so called focused proof search as an adaptation of the multi agent planning architecture, MPA, WM98] in the proof planning domain. The main component of MPA is a multi agent proof planning cell, which consists of several planning agents, a plan server, a domain server, and nally a planning cell manager. The comparison between MPA in [WM98] and our architecture is given in x7. 4.1.1 Planning ....

....of the multi agent planning architecture, MPA, WM98] in the proof planning domain. The main component of MPA is a multi agent proof planning cell, which consists of several planning agents, a plan server, a domain server, and nally a planning cell manager. The comparison between MPA in [WM98] and our architecture is given in x7. 4.1.1 Planning Agents To build up a concrete multi agent proof planning cell we choose a nite subset of agents A : fA 1 ; A n g from a given pool of planning agents. A planning agent either consists of a single reasoning system or of a composition ....

[Article contains additional citation context not shown here]

David E. Wilkins and Karen L. Myers. A Multiagent Planning Architecture. In Reid Simmons, Manuela Veloso, and Stephen Smith, editors, Proceedings of the Fourth International Conference on Articial Intelligence Planning Systems (AIPS'98), pages 154-162, Pittsburgh, PEN, USA, 7-10 1998. AAAI Press, Menlo Park, CA, USA.

....by domain analysis ] domain ] domain ] Domain domain r Analysis ] ontology ] odeling Design P S Concepts P S Models P S Methods Figure 1.1: Ontology Development Process methods into our ontology. From previous experience in building application systems [Lassila et el. 1996, Wilkins et el. 1996, Smith, 1994] from revising several different system architectures [Becker, 1993] and from previous domain analysis models [Becker and Dfaz Herrera, 1994] developed for different projects, an informal ontological model has been defined [Smith and Becker, 1997b] This model corresponds to the ....

....a ; concrete domain model can be simplified. The domain analysis techniques discussed in Chapter 2 section 2.1.7 provide some guidance on how to identify these commonalities. From our experience building scheduling systems for several different problem domains [Smith, 1987, Lassila et at. 1996, Wilkins et at. 1996, Muscettola et at. 1992] and from revising several existing knowledge based scheduling architectures [Becker, 1993, Becker and Dfaz Herrera, 1994] see Chapter 2 section 2.3, we have identified a collection of generic behaviors that can be adapted with little effort to implement applications ....

D.E. Wilkins, K.L. Myers, M. desJardins, S.F. Smith, and M.A. Becker. Multiagent planning architecture. Annual Report ECU-7150, SRI International, Menlon Park, CA, November 1996.

....and has documentation that provides more details than this report, including the MPA interface functions and interface specifications for the implemented MPA agents. A paper describing MPA was published in the proceedings of the 1998 AI Planning Systems Conference [20], and was presented at Carnegie Mellon University during the conference. 1.2 Small Unit Operations A second included report describes work done on this contract for a task added in June 1999. The task is to develop a planning and decision aide (PDA) for small unit operations (SUO) The purpose ....

....We wrote a technical report, The Air Campaign Planning Knowledge Base, describing the ACP KB, and distributed it to the ARPI community during the ISO meetings in Monterey, California in May 1998. 2 # Dr. Wilkins and Dr. Myers coauthored a paper entitled A Multiagent Planning Architecture [20], describing work on this project. Dr. Wilkins presented the paper at the AI Planning Systems conference at Carnegie Mellon University, June 7 10, 1998. He was on a panel that discussed the scalability of various AI planning approaches. We also wrote a paper for the ARPI Advanced Planning ....

[Article contains additional citation context not shown here]

David E. Wilkins and Karen L. Myers. A multiagent planning architecture. In Proc. of the

....1990) is a mature, domain independent HTN planner that uses more knowledge and has richer capabilities (such as numerical constraints and resource models) than the current DPs. Example applications include containing oil spills (Agosta Wilkins 1996) planning air campaigns for the Air Force (Wilkins Myers 1998; Lee Wilkins 1996) and joint military operations planning (Wilkins Desimone 1994) In the latter applications, the domain knowledge includes 100 to 200 operators, around 500 objects with 15 to 20 properties per object, and a few thousand initial predicate instances. Plans can include up to ....

....and new capabilities are provided, such as following plans by clicking on hyperlinks in the printed description. Constraints and Efficiency # Certain predicates are declared as functional in certain arguments, allowing a dramatic speedup, which has been documented experimentally (Myers Wilkins 1998). Functional predicates are of particular importance to reasoning about locations in planning systems, and have proven valuable In nearly every application of SIPE 2. # A sort hierarchy represents invariant properties of perpetual objects, describes the classes to which an object belongs, and ....

[Article contains additional citation context not shown here]

Wilkins, D. E., and Myers, K. L. 1998. A multiagent planning architecture. In Proc. of the 1998 International Conference on AI Planning Systems, 154--162.

No context found.

D. E. Wilkins and K. L. Myers. A Multiagent Planning Architecture. In Proc. of AIPS-98, pages 154--162, 1998.

No context found.

D. E. Wilkins and D. L. Myers. Multiagent planning architecture. In Proceedings on The Fourth International Conference on Arti cial Intelligence Planning Systems. AIPS98, June 1998.

No context found.

D. Wilkins, K. Myers, A multiagent planning architecture, in: Proc. 4th International Conference on Artificial Intelligence Planning Systems, Pittsburgh, PA, 1998, pp. 154--162.

No context found.

D.E.Wilkins, D.L.Myers. "A Multiagent Planning Architecture". Proceedings on The Fourth International Conference on Artificial Intelligence Planning Systems. AIPS'98. June710, 1998.

No context found.

D. Wilkins and K. Myers. A multiagent planning architecture. Proceedings of the Fourth International Conference on Arti cial Intelligence Planning Systems, pages 154-162, 1998.

No context found.

D.E.Wilkins, D.L.Myers. "A Multiagent Planning Architecture". Proceedings on The Fourth International Conference on Artificial Intelligence Planning Systems. AIPS'98. June7-10, 1998.

No context found.

David E. Wilkins and Karen L. Myers. A multiagent planning architecture. In AIPS-98, pages 154--162, 1998.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC