K. Decker and V. Lesser, "Generalizing the Partial Global Planning Algorithm," International Journal on Intelligent Cooperative Information Systems, vol. 1, no. 2, pp. 319--346, June 1992. [Online]. Available: http://mas.cs.umass.edu/paper/34  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of this information to get synchronized from time to time, but they cannot assume that communication is free and information can be exchanged at each moment. Most current systems rely on ad hoc heuristics (e.g. 8] or they rely on the assumption that knowledge can be shared constantly (e.g. [4, 10, 5]) Xuan et al. 18] address the problem of combining communication acts into the decision problem of a group of cooperative agents. Their framework is similar to ours but their approach is heuristic. We are interested in the most comprehensive case where cooperative agents must determine which ....

K. S. Decker and V. R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent Cooperative Information Systems, 1(2):319--346, 1992.

....finishes, and their inverses. Non computational Resource Constraints: a final type of relation is the use of physical, non computational resources. This is the major coordination relationship in some domains such as factory scheduling and office automation [36] Several works on coordination [23,8,33] have shown that only a subset of these relationships are necessary to achieve coordination. For example, 10] uses enables, facilitates and mutex to coordinate a MAS in charge of hospital patient scheduling. In this paper we represent some of the CRs described above with influence diagrams, an ....

Keith Decker and Victor Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

....constraint programming, where a set of agents or propagators try to reduce a search space to a solution, and where each propagator only sees a small set of variables. 3.3. 2 Generalized PGP The Generalized Partial Global Planning framework by Decker and Lesser also demands cooperative agents [DL95,DL92]. GPGP is a task scheduling framework. It provides a scheduler with a set of rules, called coordination mechanisms , of how it should attribute tasks to agents and in which order they should be executed. The coordination mechanisms are rules used by a central plan controller that can look into ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, pages 1(2):319-346, June 1992. ftp://ftp.cs.umass.edu/pub/lesser/decker-ijicis.ps.

....to build centralized decentralized conflict free plans. Clement [7] develops specialized agents which are responsible for HTN individual plans coordination. Several generic approaches have been proposed concerning goal decomposition, task allocation and negotiation [9] PGP [11] and later GPGP [8]) is a specialized mission representation that allows exchanges of plans among the agents. DIPART [19] is a scheme for task (re)allocation based on load balancing. Cooperation has also been treated through negotiation strategies [21] like CNP based protocols[23] or BDI approaches where agents ....

K. Decker and V. Lesser. Generalizing the partial global planning algorithm. In Int Journal of Cooperative Information Systems 92, 1992.

....Due to the peculiar structure of tasks in this domain, there arise some interesting novel questions. A prominent one is how to organise a principled approach to designing this kind of multi agent systems. Alternatives include a dominant coordination method, such as blackboard and GPGP [7] are, which mostly predetermine the overall design approach to be taken social laws, and a holistic description of the system [6] For example, one could formulate many important aspects of the discussed system by means of three highlevel social rules. The system has to keep the event data ....

Decker, K., Lesser, V.: "Generalizing the Partial Global Planning Algorithm". In Journal on Intelligent Cooperative Information Systems 1(2):319-346 (1992).

.... planner, simple body individual plan MA plan interactions between activity sequences detect resolve resource conflicts Georgeff, 1983, 15] CP DAI planner, simple body individual plan MA plan harmful (resource access) interactions plan synchronization partial global planning (PGP) 1987, [6], 8] DPS distributed vehicle monitoring DAI planner, blackboard based problem solver local plan node plan partial global plan horizontal (node node; PGP PGP) vertical (local node PGP) achieving global consistency in dynamic environments plan coordination, 1992, 29] PC office DAI ....

.... plan positive harmful interactions plan coordination protocol Table 1: Selected Single AI Multi Agent Planning Concepts As far as organizational structures and dynamic plan interactions are concerned, the approach of MAMBA primarily draws from the concept of generalized partial global planning [6]. As far as plan adaptation issues are concerned, the MAMBA approach develops on top of the plan coordination concept suggested by v. Martial [29] On this basis, the following two subsections introduce the MAMBA agent model together with the architecture of the MAMBA multi agent organization to ....

[Article contains additional citation context not shown here]

Decker, K., Lesser, V., Generalizing the Partial Global Planning Algorithm; International Journal of Intelligent & Cooperative Information Systems ;Vol. 1, No. 2; June 1992, pp. 319-346.

....in a given event processing pipeline. Among the interesting novel questions that arise as a consequence, a prominent one is how to organise a principled approach to designing this kind of multi agent systems. Alternatives include: a dominant coordination method, such as blackboard and GPGP [8], which mostly predetermine the overall design approach to be taken; an approach with an emphasis on organizational structuring: Clearly, in a mostly pipelined architecture, there are many pairs of agents that never communicate with one another at all. One could for example build an ....

Decker, K., Lesser, V.: "Generalizing the Partial Global Planning Algorithm". Journal on Intelligent Cooperative Information Systems 1(2):319--346 (1992).

....knowledge and resources, and thus differing appropriateness in solving the problem at hand. Lesser et al. Les91] describes the FA C (functionally accurate, cooperative) architecture in which agents exchange partial and tentative results in order to converge to a solution. Decker and Lesser [DL92] propose to use the generalized partial global planning algorithm as an approach to distributed coordination. This algorithm introduces non local concerns and constraints into the local scheduling of the agents activities. In [XY96] Xiangdong and Yung consider the problem of rendezvous ....

K. Decker and V. Lesser. Generalizing the Partial Global Planning Algorithm. Intl. J. Intelligent Cooperative Systems, 1(2):319--346, 1992. 122

....constraint programming, where a set of agents or propagators try to reduce a search space to a solution, and where each propagator only sees a small set of variables. 3.3. 2 Generalized PGP The Generalized Partial Global Planning framework by Decker and Lesser also demands cooperative agents [DL95,DL92]. GPGP is a task scheduling framework. It provides a scheduler with a set of rules, called coordination mechanisms , of how it should attribute tasks to agents and in which order they should be executed. The coordination mechanisms are rules used by a central plan controller that can look into ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, pages 1(2):319--346, June 1992. ftp://ftp.cs.umass.edu/pub/lesser/decker-ijicis.ps.

....agents in the same system in order to achieve global goals, or to avoid conflicts. To study different coordination approaches, different models have been studied, including negotiation [Smi78, Dur88, KW91, ZR93, RZ94] economics [MFH88, Wel92, Kra93] social laws [GR94, TM89, ST92] and others [DL92, ER91, GK93] One way used by agent designers to quantify the agents performance is to let the agents compute a utility function. In certain systems, this function represents the gain of the agent from choosing an action because of the existence of other agents, in contrast to the cost incurred ....

K. S. Decker and V. R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent Cooperative Information Systems, 1(2):319-- 346, 1992.

....have more freedom to move (either physically or abstractly) the problem of scaling up is more significant. Nevertheless, Lesser et al. also considered problems of scale up and developed methods to reduce the communication needed to reach globally consistent solutions. For example, the GPGP model [8] extended the PGP ideas by allowing more agent heterogeneity, the exchange of more truly partially global information at multiple levels of abstraction and the use of separate scheduling algorithms. However, GPGP provides a meta level for coordinated planning and is not a coordination algorithm in ....

Keith Decker and Victor Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent Cooperative Information Systems, 1(2):319--346.

....aim to solve a domain problem. Therefore, they might not achieve the optimal solutions since the world might change rapidly and they might have limited resources. Instead, the agents have to use all the information available to them to cooperate in achieving acceptable solutions. More recently [DL92] the Partial Global Planning technique has been extended to let the agents transmit more abstract information among themselves, and detect coordination relationships that can be useful to improve the performance of the agents. The distinct kinds of coordination relationships the agents detect ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent Cooperative Information Systems, 1(2):319--346, 1992.

.... Georgeff s early work on multiagent planning assumed that there was no basic conflict among agent goals, and that coordination was all that was necessary to guarantee success (Georgeff, 1983, 1984; Stuart, 1985) Similarly, planning in the context of Lesser, Corkill, Durfee, and Decker s research (Decker Lesser, 1992, 1993b, 1993a) often involves coordination of activities (e.g. sensor network computations) among agents who have no inherent conflict with one another (though surface conflict may exist) Planning here means avoidance of redundant or distracting activity, efficient exploration of the search ....

....global goal. Our agents, on the other hand, are motivated to achieve individual goals. Second, unlike our formal approach to mechanism design, Lesser s work has historically been heuristic and experimental, although his more recent work has explored the theoretical basis for system level phenomena (Decker Lesser, 1992, 1993a, 1993b) Sycara has examined a model of negotiation that combines case based reasoning and optimization of multi attribute utilities (Sycara, 1988, 1989) In particular, while we assume that agents goals are fixed during the negotiation, Sycara is specifically interested in how agents can ....

Decker, K. S., & Lesser, V. R. (1992). Generalizing the partial global planning algorithm.

No context found.

K. Decker and V. R. Lesser, "Generalizing the Partial Global Planning Algorithm", International Journal of Intelligent and Cooperative Information Systems, 1(2), 1992.

....respecting global deadlines on those tasks. The testbed was developed to explore di erent types of coordination protocols and compare them. The goal was to compare the performance of specialized coordination protocols (such as seen in [17] against generic protocols (like Contract Net[1] and GPGP[8]) We hoped to quantitatively evaluate how these techniques functioned in the environment, in terms of time to converge, the quality and stability of the resulting organization, and the time, processing and message costs. JAF and T ms were used extensively, to develop the agents and model their ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319-346, June 1992.

....leaving the implementation to the developer beyond a few API conventions. Much of the generality available in existing JAF components is derived from their common use of a powerful, domain independent representation of how agents can satisfy di erent goals. This representation, called T ms [4, 5], allows complex interactions to be phrased in a common language, allowing individual components to interact without having direct knowledge of how other components function. Implemented components in JAF are designed to operate with relative autonomy. Coincidentally, a reasonable analogy for a ....

Keith Decker and Victor Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent Cooperative Information Systems, 1(2):319{ 346, 1992.

No context found.

Decker, K. and Lesser, V., "Generalizing The Partial Global Planning Algorithm," International Journal on Intelligent Cooperative Information Systems, 1(2):319-346, 1992.

....a quantitative theory, qualitative statements may be derived from it. We are now focusing on generalizing the partial global planning mechanism. This process involves identifying the types of coordination relationships that are used by the basic PGP algorithm and that exist but that are not used [6], developing a conceptual model that clearly specifies the roles of a PGP style coordination algorithm as identifying coordination relationships and producing behaviors (primarily the creation and refinement of local scheduling constraints) and generalizing the partial global planning algorithm ....

.... a conceptual model that clearly specifies the roles of a PGP style coordination algorithm as identifying coordination relationships and producing behaviors (primarily the creation and refinement of local scheduling constraints) and generalizing the partial global planning algorithm itself (GPGP) [6]. Our current approach views the coordination mechanism as modulating local control, not supplanting it a two level process that makes a clear distinction between coordination behavior and local scheduling [3] By concentrating on the creation of local scheduling constraints, we avoid the ....

[Article contains additional citation context not shown here]

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2), June 1992.

....scheduling. In this work no limitations were placed on interactions (thus necessitating heuristic scheduling) and problems were divided (usually with some overlap) among multiple agents. Coordination between agents was handled using the Generalized Partial Global Planning (GPGP) approach[Decker and Lesser, 1992] . At the request of the coordination algorithm, agents would make commitments to completing particular methods by particular times. The heuristic scheduling in this work was done using an early version of the basic algorithm described in this paper. An example of a problem to be solved by the ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

....situations is also used when the system does not have a sufficiently rich domain model to fully explain the failure. Our approach to learning coordination rules can be couched in the following more general perspective that comes out of work by Decker and Lesser on generic coordination strategies [3]. Each agent makes scheduling decisions based on a subjective view of its own and other agents activities. In certain situations this subjective view will lead to ineffective inappropriate problem solving because certain local and non local task interrelationships have not been appropriately 2 ....

....plan Get RTT between Both Ends to another node, this node may have a different viewpoint and find other plans and tasks that it decides are more important than the proposed plan in a similar problem. In such a situation, a network control regime similar to generic partial global planning (GPGP) [3] is required. Even if the local priority of a plan is low, if other agents request the plan and its result facilitates their reasoning, it should be performed. How to decide plan priorities, especially for plans requested by other agents, is one of the difficult issues for real world systems. An ....

K. S. Decker and V. R. Lesser, "Generalizing the Partial Global Planning Algorithm," Int. Jour. of Intelligent Cooperative Information Systems, Vol. 1, No. 2, pp. 319-346, 1992.

....are the most important part of the TMS framework, since they supply most of the characteristics that make one task environment unique and different from another. Typically a model will define different classes of effects, such as causes, facilitates, cancels, constrains, inhibits, and enables [10]. This section contains definitions for four common classes of effects that have been useful in modeling different environments. When non local effects occur between methods associated with different agents, we call them coordination relationships [9, 10] Enables. If task T a enables method M , ....

....cancels, constrains, inhibits, and enables [10] This section contains definitions for four common classes of effects that have been useful in modeling different environments. When non local effects occur between methods associated with different agents, we call them coordination relationships [9, 10]. Enables. If task T a enables method M , then the maximum quality q(M; t) 0 until T a is completed and the result is available, when the maximum quality will change to the initial maximum quality q(M; t) q 0 (M) enables(T a ; M; t; d; q; 1; 0] t Theta(T a ; d 0 (M) q 0 ....

[Article contains additional citation context not shown here]

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2), June 1992.

....respecting global deadlines on those tasks. The testbed was developed to explore different types of coordination protocols and compare them. The specific idea was to compare the performance of specialized protocols (such as seen in [11] against generic protocols (like Contract Net[1] and GPGP[3]) MASS was used to build a regular day in the house it simulates the tasks requested by the occupants, maintains the status of all environmental resources, simulates agent interactions with the house and resources, and manages sensors available to the agents. Using MASS features, we can ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

....return to contingency analysis in Section 3. While Design to Criteria is a research focus in its own right, it is often incorporated into other research projects or used as an analysis expert by other tools. For example, in multi agent systems research, Design toCriteria is coupled with the GPGP [5] coordination module enabling an agent to coordinate its activities with the activities of other agents. GPGP operates by exchanging the agents local views, detecting task interactions, then forming commitments over the interactions to handle temporal sequencing of activities. GPGP modulates ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

....are the most important part of the T MS framework, since they supply most of the characteristics that make one task environment unique and different from another. Typically a model will define different classes of effects, such as causes, facilitates, cancels, constrains, inhibits, and enables (Decker Lesser 1992). This section contains a definition for one common class of effect that has been useful in modeling different environments; another definition will be presented in Section 16.1.4; see Decker Lesser (1993e) for a more complete set of definitions. When non local effects occur between methods ....

....different environments; another definition will be presented in Section 16.1.4; see Decker Lesser (1993e) for a more complete set of definitions. When non local effects occur between methods associated with different agents, we call them coordination relationships (Decker Lesser 1993a, Decker Lesser 1992). An important effect, used by the Partial Global Planning (PGP) algorithm (see Chapter 8 and Durfee Lesser (1991) but never formally defined, is the facilitates effect. Computationally, facilitation occurs when information from one task, often in the form of constraints, is provided that ....

Decker, K. S. & Lesser, V. R. (1992), `Generalizing the partial global planning algorithm ', International Journal of Intelligent and Cooperative Information Systems 1(2), 319--346.

....In a multi agent system, each agent can only have a partial view of other agents behavior. Therefore, in order to coordinate the agents activities, the agents need to have a mechanism to bridge their activities based on the partial knowledge. Commitments has emerged, among many research groups [1 3, 8], as the bridge for multi agent coordination and planning. By definition, a commitment specifies a pledge to do a certain course of action [9] A number of commitment semantics have been proposed, for example, the Deadline commitment C(T ; Q; t dl ) in [3] means a commitment to do (achieve ....

....among many research groups [1 3, 8] as the bridge for multi agent coordination and planning. By definition, a commitment specifies a pledge to do a certain course of action [9] A number of commitment semantics have been proposed, for example, the Deadline commitment C(T ; Q; t dl ) in [3], means a commitment to do (achieve quality Q or above for) a task T at a time t so that it finishes before a specified deadline, t dl . When such a pledge is offered, the receiving agent can then do its own reasoning and planning based on this commitment, and thus achieves coordination between ....

[Article contains additional citation context not shown here]

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. In International Journal of Intelligent and Cooperative Information Systems. 1(2):319-346, 1992.

....goal relationships. KSI invalidation uses knowledge about supergoal and subgoal relationships to understand the effect of KSI executions on other KSIs goals. There are four general categories of goal relationships that can be used (via KSI rating heuristic functions) to schedule domain KSIs [9]: Domain Relations: This set of relations is generic in that they apply to multiple domains and domain dependent in that they can be evaluated only with respect to a particular domain. Examples of domain relations include inhibits, cancels, constrains, facilitates, causes, enables, and ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2), June 1992.

....a quantitative theory, qualitative statements may be derived from it. We are now focusing on generalizing the partial global planning mechanism. This process involves identifying the types of coordination relationships that are used by the basic PGP algorithm and that exist but that are not used (Decker and Lesser, 1992), developing a conceptual model that clearly specifies the roles of a PGP style coordination algorithm as identifying coordination relationships and producing behaviors (primarily the creation and refinement of local scheduling constraints) and generalizing the partial global planning algorithm ....

.... a conceptual model that clearly specifies the roles of a PGP style coordination algorithm as identifying coordination relationships and producing behaviors (primarily the creation and refinement of local scheduling constraints) and generalizing the partial global planning algorithm itself (GPGP) (Decker and Lesser, 1992). Our current approach views the coordination mechanism as modulating local control, not supplanting it a two level process that makes a clear distinction between coordination behavior and local scheduling (Corkill and Lesser, 1983) By concentrating on the creation of local scheduling ....

[Article contains additional citation context not shown here]

Decker, K. S. and Lesser, V. R. (1992). Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346.

....of physical locations (such as the main TidBits archive ftp.tidbits.com or its various mirrors) 2. 1 Subproblem Interaction Various kinds of goal interrelationships like facilitates, enables, overlaps, and subsumes that exist between subproblems can be detected and exploited in a variety of ways [5, 6]. For example, uncertainty that may arise from incomplete local information can be reduced through detection and subsequent exploitation of overlaps and subsumes interrelationships. However, the process of detecting and exploiting goal interrelationships involves providing the agent with a more ....

....schedules (what actions to take, in what order, and at what point in time) and an execution monitor that oversees action execution to anticipate problems and slipped deadlines. This architecture is based on the GPGP (Generalized Partial Global Planning) architecture described in detail in [5, 6]. For example, upon receiving its task Get Online Reviews , the Online Reviews agent calls a planner to expand the task into a query plan consisting of primitive actions like Use FTP and Use Uncover . A real time scheduler is invoked in conjunction with coordination mechanisms to enable the ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, jun 1992.

....interrelationships between sibling goals at various levels of the tree so that, for example, solving one goal before another goal can facilitate the later goal s solution quality. The goal interrelationships can be of various types such as facilitates, enables, overlaps, hinders, favors, and so on[9, 11, 26]. In terms of our goal representation, a facilitates interrelationship implies that the values of a solution output parameter of the facilitating goal can, in some way, determine an optional goal specification parameter of the facilitated goal. The facilitated goal could have pursued its activity ....

....is the explicit recognition and exploitation or avoidance of redundancy, leading to increased robustness or decreased resource demands depending on the context and the structure of the domain. We now discuss implemented aspects of FA C that have direct relevance to the CIG task. Decker and Lesser[9, 10, 11] provide detailed studies of quantitative trade offs involved in explicit recognition and exploitation of task interrelationships for use in multi agent coordination. Von Martial s work[26] on coordination in multi agent planning using favors goal interrelationships and temporal interactions is ....

K. S. Decker and V. R. Lesser, "Generalizing the Partial Global Planning Algorithm", International Journal of Intelligent and Cooperative Information Systems 1(2), June 1992, pp 319-346.

....and coordination algorithms will describe when and where communication actually occurs (see communication actions in Section 5.3, and the concept of agency in Section 5. 1) model will define different classes of effects, such as causes, facilitates, cancels, constrains, inhibits, and enables [10]. This section contains definitions for four common classes of effects that have been useful in modeling different environments. When non local effects occur between methods associated with different agents, we call them coordination relationships [11, 10] Enables. If task T a enables method M , ....

....cancels, constrains, inhibits, and enables [10] This section contains definitions for four common classes of effects that have been useful in modeling different environments. When non local effects occur between methods associated with different agents, we call them coordination relationships [11, 10]. Enables. If task T a enables method M , then the maximum quality q(M; t) 0 until T a is completed and the result is available, when the maximum quality will change to the initial maximum quality q(M; t) q 0 (M) Another way to view this effect is that it changes the earliest start time ....

[Article contains additional citation context not shown here]

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

....of other agents behavior. Therefore, in order to coordinate the agents activities, the agents need to have a mechanism to bridge their activities based on the partial knowledge. Commitments has emerged, among many research groups [Cohen and Levesque, 1990, Castelfranchi 1993, Jennings 1993, Decker and Lesser, 1992] as the bridge for multi agent coordination. By definition, a commitment specifies a pledge to do a certain course of action [Jennings 1996] A number of commitment semantics have been proposed, for example, the Deadline commitment C(T ; Q; t dl ) in [Decker and Lesser, 1992] means a ....

....1993, Jennings 1993, Decker and Lesser, 1992] as the bridge for multi agent coordination. By definition, a commitment specifies a pledge to do a certain course of action [Jennings 1996] A number of commitment semantics have been proposed, for example, the Deadline commitment C(T ; Q; t dl ) in [Decker and Lesser, 1992], means a commitment to do (achieve quality Q or above for) a task T at a time t so that it finishes before a specified Effort sponsored by the Defense Advanced Research Projects Agency (DARPA) and Air Force Research Laboratory Air Force Materiel Command, USAF, under agreement number ....

[Article contains additional citation context not shown here]

Keith S. Decker and Victor R. Lesser, "Generalizing the partial global planning algorithm." International Journal of Intelligent and Cooperative Information Systems. 1(2):319-346. June 1992.

....and Lesser, 1994] In the rest of this paper we will concentrate only on the types of communication each module produces. The five modules described in [Decker and Lesser, 1994] form a basic set that provides similar functionality to the original partial global planning algorithm as explained in [Decker and Lesser, 1992] . All modules rest on a shared substrate that acts as the agent s central decision maker, deciding on which plan schedule to follow and when to terminate processing on a problem. The shared substrate also handles retractions of commitments to other agents when circumstances change and the agent ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319-- 346, June 1992.

.... the Distributed Airport Resource Management System (DIS ARM) 2 Related Work The use of meta level information to define the interactions between agents has been studied extensively by Durfee and Lesser via the use of partial global plans [5] This work has been extended by Decker and Lesser [3, 4] to incorporate more sophisticated coordination relationships. According to this framework, we can view our detection of potential loan requests using texture measures to be an identification of facilitating relationships, and our modification of the scheduling algorithm as an attempt to exploit ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

.... his staff example) or within portions controllable by multiple agents (as in the case of the overlapping tasks in the JFACC example) 4 The five mechanisms we described in [9] form a basic set that provides similar functionality to the original partial global planning algorithm as explained in [6]. Mechanism 1 exchanges useful private views of task structures; Mechanism 2 communicates results; Mechanism 3 handles redundant methods; Mechanisms 4 and 5 handle hard and soft coordination relationships. More mechanisms can be added, such as a load balancing mechanism. The mechanisms are ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

No context found.

K. Decker and V. R. Lesser, "Generalizing the Partial Global Planning Algorithm", International Journal of Intelligent and Cooperative Information Systems, 1(2), 1992.

.... on an individual task is a function of the timing and choice of agent actions ( local effects ) and possibly previous task executions ( non local effects ) When local or non local effects exist between tasks that are known by more than one agent, they become coordination relationships [4, 3]. The basic purpose of the objective model is to formally specify how the execution and timing of tasks affect quality and duration. At the lowest level, each method (leaf task) M at time t can produce, if executed, some maximum quality q(M; t) in some amount of time d(M; t) each method has an ....

....The addition of correlated noise in the environment can then cause these new, more complex coordination mechanisms to break down, producing the phenomenon recognized as distraction. In the long term, we are working towards a complete characterization of generalized partial global planning[3] as a first step towards a theory of coordination in distributed problem solving. We and other researchers are also using our task structure characterization for the analysis and simulation of problems in real time and parallel scheduling. ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

No context found.

K. Decker and V. R. Lesser, "Generalizing the Partial Global Planning Algorithm", International Journal of Intelligent and Cooperative Information Systems, 1(2), 1992, pp 319 -- 346.

....together an efficient team on the fly, and the ability to intelligently and dynamically assign and track these agent resources. 2. 2 The DECAF Agent Architecture This architecture is based on the one used to test the GPGP (Generalized Partial Global Planning) coordination approach described in [4, 5]. In the context of Figure 2, for example, upon receiving its task Get Online Reviews, the Online Reviews agent calls a planner to expand the task into a query plan consisting of primitive actions like Use FTP and Use Uncover. A real time scheduler is invoked in conjunction with ....

....of new local scheduling constraints, called commitments, which can be communicated to other agents. Various kinds of subproblem interaction or goal interrelationships like facilitates, enables, overlaps, and subsumes that exist between subproblems can be detected and exploited in a variety of ways [4, 5]. Different coordination mechanisms are used to deal with different classes of interdependencies. Some example mechanisms we have developed include: ffl a mechanism that recognizes and responds to tasks that have an impact on other agents but are not believed to be known by the other agents ffl a ....

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, jun 1992.

....sources. First and foremost is our own and others work in the DVMT and similar domain environment simulators [Corkill and Lesser, 1983, Lesser and Corkill, 1983, Durfee and Lesser, 1987, Durfee et al. 1987, Cohen et al. 1989, Decker et al. 1990, Carver et al. 1991, Decker et al. 1991, Decker et al. 1992] . It is from this work that the basic model form the execution of interrelated tasks is taken. One possible form might have been to create a simulator (ala Tileworld [Pollack and Ringuette, 1990] but it would be very hard to state good featural characterizations using a simulator. The ....

.... the PGP line of research, where the exchange of meta level information about the schedules of remote node was used, for example, to avoid redundant work [Durfee, 1987] The work that has followed this has emphasized the use of abstraction to reduce communication and detect activity interactions [Decker and Lesser, 1992, Durfee and Montgomery, 1990] Adaptation: Self organization of multi agent systems is another coordination behavior. The Contract Net, and other related market based organizations, represents one style of self organization as agents create temporary organizational structures through ....

[Article contains additional citation context not shown here]

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

....for representing coordination problems, and summarize the assumptions we make about an agent s internal architecture. We then describe the GPGP substrate and five coordination mechanisms. 1 Previous work has shown how the GPGP approach can duplicate and extend the behaviors of the PGP algorithm [5]; Section 4 summarizes several new results that are reported in [4] concerning this approach s performance, adaptability, and extendibility. We conclude with a look at our future directions. 1.1 Representing The Task Environment Coordination is the process of managing interdependencies between ....

....of the subjective task structure of the episode or in the form of local and non local commitments to tasks in the task structure. The five mechanisms we will describe in this paper form a basic set that provides similar functionality to the original Partial Global Planning algorithm as shown in [5]. Mechanism 1 exchanges useful private views of task structures; Mechanism 2 communicates results; Mechanism 3 handles redundant methods; Mechanisms 4 and 5 handle hard and soft coordination relationships. More mechanisms can be added, such as one to update utilities across agents as discussed in ....

[Article contains additional citation context not shown here]

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

No context found.

Decker, Keith S. and Lesser, Victor R. 1992. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems 1(2):319-- 346.

....interrelationships between sibling goals at various levels of the tree so that, for example, solving one goal before another goal can facilitate the later goal s solution quality. The goal interrelationships can be of various types such as facilitates, enables, overlaps, hinders, favors, and so on[7, 20]. In terms of our goal representation, a facilitates interrelationship implies that the values of a solution output parameter of the facilitating goal can, in some way, determine an optional goal specification parameter of the facilitated goal. The facilitated goal could have pursued its activity ....

....is the explicit recognition and exploitation or avoidance of redundancy, leading to increased robustness or decreased resource demands depending on the context and the structure of the domain. We now discuss implemented aspects of FA C that have direct relevance to the CIG task. Decker and Lesser[7, 8] provide detailed studies of quantitative trade offs involved in explicit recognition and exploitation of task interrelationships for use in multi agent coordination. Von Martial s work[20] on coordination in multi agent planning using favors goal interrelationships and temporal interactions is ....

K. S. Decker and V. R. Lesser, "Generalizing the Partial Global Planning Algorithm", International Journal of Intelligent and Cooperative Information Systems 1(2), June 1992, pp 319-346.

....agent detect interrelationships between sibling goals at various levels of the tree so that, for example, solving one goal before another goal can facilitate the latter goal s solution quality. The goal interrelationships can be of various types such as facilitates, enables, overlaps, and hinders[Decker and Lesser, 1992, Decker and Lesser, 1993a, Decker, 1994] A facilitates interrelationship implies that the values of a solution output parameter of the facilitating goal can, in some way, determine an optional goal specification parameter of the facilitated goal. The facilitated goal could have been pursued ....

....have to execute and the overall utility of the system as a whole. Knowing the task environment better has a direct impact on an agent s ability to realize the implications of its decisions and their non local effects. Later in the thesis, we discuss in more detail 10 how, in the GPGP context[Decker and Lesser, 1992, Decker and Lesser, 1995] learning task environment characteristics can be achieved. ffl How to enable individual agents to represent and reason about actions, plans and knowledge of other agents in order to coordinate with them. The ability to model another agent s goals and beliefs has a ....

[Article contains additional citation context not shown here]

Decker, K. S. and Lesser, V. R. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1(2):319--346, June 1992.

No context found.

K. Decker and V. Lesser, "Generalizing the Partial Global Planning Algorithm," International Journal on Intelligent Cooperative Information Systems, vol. 1, no. 2, pp. 319--346, June 1992. [Online]. Available: http://mas.cs.umass.edu/paper/34

No context found.

Decker K. S. & Lesser V. R. "Generalizing the Partial Global Planning Algorithm". International Journal of Intelligent and Cooperative Information Systems, 1(2), p.319-346, 1992.

No context found.

Decker, K. S. Lesser, V. R., 1992. "Generalizing the partial global planning algorithm." Int. Journal of Intelligent and Cooperative Information Systems, 1 (2).

No context found.

Decker, K. S. Lesser, V. R., 1992. "Generalizing the partial global planning algorithm." Int. Journal of Intelligent and Cooperative Information Systems, 1 (2).

No context found.

K. S. Decker and V. R. Lesser. Generalizing the partial global planning algorithm. Int. J. of Intelligent and Cooperative Information Systems, 1(2):319 346, June 1992.

No context found.

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent Cooperative Information Systems, 1(2):319-- 346, 1992.

No context found.

Keith S. Decker and Victor R. Lesser. Generalizing the partial global planning algorithm. International Journal of Intelligent and Cooperative Information Systems, 1:319--346, 1992.

First 50 documents  Next 50

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