S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In [16], pages 228--235, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....directed dynamical behavior. This can also be related to works on the definition of generic robot controllers [10, 32] and to task level robot programming models [31] ffl execution models for planning in articifical intelligence, especially reactive planning, or more acurately reactive plans [34], and temporal planning [30] an application of this can be the execution of high level reactive plans on robotic architectures; it relates to the definition of hierarchical architectures for autonomous systems [25] and the automatic synthesis of sequencings [16] ....

S. Thi'ebaux, J. Hertzberg. A Semi-Reactive Planner Based on a Possible Models Action Formalization. In Proceedings of the First International Conference on Artificial Intelligence Planning Systems (AIPS92), 228235, Morgan Kaufmann, 1992.

....approach is presented, and it is shown that this extended approach is essentially equivalent for disjunctive 22 planning problems to the linear connection method employing the linearity criterion proposed in this paper. Our approach is related to others which generate disjunctive plans such as [TH92] Here, the situations succeeding a situation roughly correspond to the possible models. But there are notable differences. Our situation descriptions are multisets instead of sets with the obvious advantage for modelling resources. The second part of the paper dealt with the efficiency of the ....

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In J. Hendler, editor, Artificial Intelligence Planning Systems: Proceedings of the First International Conference (AIPS92), San Mateo, CA, 1992. Morgan Kaufmann.

....bound the temporal extent of constraints and default persistence. From a computational point of view, within this framework it is forward temporal prediction tasks that currently offer the most promise, due to recent work on state based update approaches [Chou and Winslett, 1993; del Val, 1993b; Thi ebaux and Hertzberg, 1992]. As already mentioned, there is a very tight correspondence between the class of theories of action of this paper and state based approaches. Furthermore, it is worth noting that the kinds of persistence policy we have suggested and that we believe are most likely to be useful in practice are for ....

Thi'ebaux, Sylvie and Hertzberg, Joachim 1992. A semi-reactive planner based on a possible models action formalization. In Proceedings of the First International Conference on Artificial Intelligence Planning Systems.

....by proving this specification formula. Frequently this process is formulated in some non classical logic formalism, for example linear logic [2, 5, 12] or temporal logic [18] or is based on some logical action theory. In the other approach, the planning process is carried out in an algorithmic way [19, 7, 6, 20]. Our approach to deductive planning is situated between these two groups. As in all deductive approaches, we start from a logical action theory, i.e. the world and the actions are specified in logic. From this logical specification, we automatically construct an action graph using a theorem ....

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In AI Planning Systems, Proceedings of the First International Conference, pages 228--235. Morgan Kaufmann, 1992.

.... are also used for a broad class of planning applications, e.g. 28] It is useful to combine work on these two topics, the perspective being to build planners that cope with uncertainty, and incrementally increase the plan quality if time permits, e.g. 10, 1] We have earlier described pascale [23], a system for planning under uncertainty, based on a theory of actions with uncertain outcomes [3] which deals explicitly with uncertainty arising from the incompleteness or ambiguity of information. pascale s representation of reaction plans allows them to be generated from first principles, as ....

....generated from first principles if time permits, and that, third, can be easily reused or incrementally extended if time is tight. We do not yet exploit probabilistic information or deal with anytime planning issues; this will be done in the next section. In fact, we just summarize key issues from [23] here. A plan is a bipartite directed graph with two types of nodes: T nodes representing tasks, i.e. occurrences of actions in a plan, and M nodes representing possible models. This is to be interpreted as follows: a T node T preceded by some M node M means that the plan specifies that T is ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In [16], pages 228--235, 1992.

.... INSTITUTE OF TECHNOLOGY MELBOURNE, FLORIDA ANYTIME REACTION PLANNING IN PROBABILISTIC LOGIC a thesis submitted to the department of computer science of the florida institute of technology in partial fulfillment of the requirements for the degree of master of science By Sylvie Thi ebaux June 1992 c fl Copyright 1992 by Sylvie Thi ebaux ii ANYTIME REACTION PLANNING IN PROBABILISTIC LOGIC June 1992 Dr. Mordechay Schneider Associate Professor Department of Computer Science Florida Institute of Technology (Principal Advisor) Dr. Laurene Fausett Associate Professor Department ....

....that purpose, some default behavior must be encoded in a library that might be extended to include better plans newly generated. The system should propose a compromise between the quality of the plans generated and the bounded time resources available. Introduction 3 PASCALE [Thi ebaux, 1991; Thi ebaux and Hertzberg, 1992] , the system we propose to improve, is an attempt to maintain the unified view of a planner for unpredictable domains, by combining ideas from two of the individual areas, namely formalization and reactivity. PASCALE is a semi reactive planner, i.e. a planner that is able to incrementally ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In J. Hendler, editor, Artificial Intelligence Planning Systems: Proc. First International Conference (AIPS92), San Mateo, CA, June 1992. Morgan Kaufmann.

....in general (left column) the respective instances of these steps for the possible worlds formalism in particular (right column) and the respective sections in this paper that deal with the respective steps. Single aspects of this work have been reported in [ Brewka and Hertzberg, to appear; Thi ebaux and Hertzberg, 1992; Thi ebaux et al. 1993 ] from which texts this paper borrows occasionally. 2 The Action Formalism Under the hypothesis (1) we could carry out our case study using any reasonable action formalism as presented, e.g. in [ Sandewall, 1991 ] As mentioned, the required expressivity of the ....

....only part of the demonstration substrate in our case study for planner construction, we do not go into detail concerning the underlying execution model and the possibilities to use them for execution monitoring, replanning, or reusing old plans. The interested reader may find hints about this in [ Thi ebaux and Hertzberg, 1992 ] The example plan Pi c is rather exceptional, given the simple plan definition that we have: Pi c makes sense in the sense that it correctly models the postconditions of all actions labeling its T nodes, as applied in their respective predecessor worlds. The plan definition does not enforce ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In J. Hendler, editor, Artificial Intelligence Planning Systems: Proceedings of the First International Conference (AIPS92), pages 228--235, Morgan Kaufmann, San Mateo, CA, 1992.

.... (left column) the respective instances of these steps for the possible worlds formalism in particular (right column) and the respective sections in this paper that deal with the respective steps (middle column) Single aspects of this work have been reported in [ Brewka and Hertzberg, in press; Thi ebaux and Hertzberg, 1992; Thi ebaux et al. 1993 ] from which texts this paper borrows occasionally. 2 The Action Formalism Under the hypothesis (1) we could carry out our case study using any reasonable action formalism as presented, e.g. in [ Sandewall, 1991 ] As mentioned, the required expressivity of the ....

....only part of the demonstration substrate in our case study for planner construction, we do not go into detail concerning the underlying execution model and the possibilities to use them for execution monitoring, replanning, or reusing old plans. The interested reader may find hints about this in [ Thi ebaux and Hertzberg, 1992 ] The example plan Pi c is rather exceptional, given the simple plan definition that we have: Pi c is meaningful in the sense that it correctly models the postconditions of all actions labeling its T nodes, as applied in their respective predecessor worlds. The plan definition does not ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In J. Hendler, editor, Artificial Intelligence Planning Systems: Proceedings of the First International Conference (AIPS92), pages 228--235, San Mateo, CA, 1992. Morgan Kaufmann.

....Obviously, it is useful to combine work on these two topics, the perspective being to build planners that cope with uncertainty, and incrementally increase the plan quality if time permits. Examples of such works are [ Drummond and Bresina, 1990; Beetz and McDermott, 1992 ] In a previous paper [ Thi ebaux and Hertzberg, 1992 ] we have described pascale, a system for planning under uncertainty. This system is based on a particular theory of actions with uncertain outcomes [ Brewka and Hertzberg, To appear ] which deals explicitly with uncertainty arising from the incompleteness or ambiguity of information. ....

....that all the forms of uncertainty actually occur in every planning application. It enables a whole spectrum of implementations; special cases being, e.g. 2 2 FORMALIZING ACTIONS IN PROBABILISTIC LOGIC classical linear plans and universal plans [ Schoppers, 1987 ] The paper is based on [ Thi ebaux, 1992 ] which contains additional details. It is organized as follows. Section 2 describes our formalization of actions which blends our previously used formalism with Nilsson s [ 1986 ] probabilistic logic. Section 3 deals with the representation of reaction plans and with its advantages for ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In [Hendler, 1992] , pages 228--235, 1992.

.... (left column) the respective instances of these steps for the possible worlds formalism in particular (right column) and the respective sections in this paper that deal with the respective steps (middle column) Single aspects of this work have been reported in [ Brewka and Hertzberg, in press; Thi ebaux and Hertzberg, 1992; Thi ebaux et al. 1993 ] from which texts this paper borrows occasionally. 2 The Action Formalism Under the hypothesis (1) we could carry out our case study using any reasonable action formalism as presented, e.g. in [ Sandewall, 1991 ] As mentioned, the required expressivity 3 planner ....

....only part of the demonstration substrate in our case study for planner construction, we do not go into detail concerning the underlying execution model and the possibilities to use them for execution monitoring, replanning, or reusing old plans. The interested reader may find hints about this in [ Thi ebaux and Hertzberg, 1992 ] The example plan Pi c is rather exceptional, given the simple plan definition that we have: Pi c is meaningful in the sense that it correctly models the postconditions of all actions 11 start table2up back2up spin bk ry fd up of ot bk ry fd up of ot bk ry fd up of ot ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In J. Hendler, editor, Artificial Intelligence Planning Systems: Proceedings of the First International Conference (AIPS92), pages 228--235, San Mateo, CA, 1992. Morgan Kaufmann.

....Obviously, it is useful to combine work on these two topics, the perspective being to build planners that cope with uncertainty, and incrementally increase the plan quality if time permits. Examples of such works are [Drummond and Bresina, 1990; Beetz and McDermott, 1992] In a previous paper [Thi ebaux and Hertzberg, 1992], we have described pascale, a system for planning under uncertainty. This system is based on a particular theory of actions with uncertain outcomes [Brewka and Hertzberg, To appear] which deals explicitly with uncertainty arising from the incompleteness or ambiguity of information. pascale s ....

....theory. Note that the framework does not presuppose that all the forms of uncertainty actually occur in every planning application. It enables a whole spectrum of implementations; special cases being, e.g. classical linear plans and universal plans [Schoppers, 1987] The paper is based on [Thi ebaux, 1992], which contains additional details. It is organized as follows. Section 2 describes our formalization of actions which blends our previously used formalism with Nilsson s [1986] probabilistic logic. Section 3 deals with the representation of reaction plans and with its advantages for planning ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In [Hendler, 1992], pages 228--235, 1992.

.... into a planner in general (left column) the respective instances of these steps for the possible worlds formalism in particular (right column) and the respective sections in this paper that deal with the respective steps (middle column) Single aspects of this work have been reported in [3, 23, 24], from which texts this paper borrows occasionally. Note that the formalism from [3] is only used as a tool here. Consequently, discussing its strengths and weaknesses or relating it to comparable formalisms is out of the scope of this paper, and we refer to the original article. 2 The Action ....

....only part of the demonstration substrate in our case study for planner construction, we do not go into detail concerning the underlying execution model and the possibilities to use them for execution monitoring, replanning, or reusing old plans. The interested reader may find hints about this in [23]. The example plan Pi c is rather exceptional, given the simple plan definition that we have: Pi c is meaningful in the sense that it correctly models the postconditions of all actions labeling its T nodes, as applied in their respective predecessor worlds. The plan definition does not ....

[Article contains additional citation context not shown here]

S. Thi'ebaux and J. Hertzberg. A semi-reactive planner based on a possible models action formalization. In J. Hendler, editor, Artificial Intelligence Planning Systems: Proceedings of the First International Conference (AIPS92), pages 228--235, San Mateo, CA, 1992. Morgan Kaufmann.

No context found.

Sylvie Thi'ebaux and J. Hertzberg 1992 A semi-reactive planner based on a possible models action formalization. In In James Hendler (ed.), Proc. First Int. Conf. on AI Planning Systems, San Mateo: Morgan Kaufmann, pp. 228--235

No context found.

Sylvie Thi'ebaux and J. Hertzberg 1992 A semi-reactive planner based on a possible models action formalization. In In James Hendler (ed.), Proc. First Int. Conf. on AI Planning Systems, San Mateo: Morgan Kaufmann, pp. 228--235

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