Subbarao Kambhampati, Amol Mali & Biplav Srivastava, Hybrid planning in partially hierarchical do- mains, Proceedings of the National Conference on Arti cial Intelligence (AAAI), 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....a starting point for the implementation of a planning system. HTN based planning, also known as task decomposition planning, is among the oldest approaches for providing domainspecific knowledge to a planning system. While in the generic case HTN planning may be faced with practical difficulties [21], this approach is considered appropriate for knowledge rich domains, which can provide applications specific knowledge to assist plan generation [22] Interactive Storytelling constitutes such a knowledge rich application, not least because of the authoring process involved in the description of ....

Kambhampati, S., Mali, A., and Srivastava, B, 1998. Hybrid planning for partially hierarchical domains. In Proceedings of AAAI-98.

....Finding out the better search and rescue strategies for large scale disaster involves the state of the art planning and multi agent research, such as teamwork [Tambe and Zhang. 1998] planning under uncertaintity [Pollack, 1998] resource boundedness[Russel, 1995] hierarchical planning [ Kambhampati et al. 1998 ] and real time planning. While we cannot possibly create an exhaustive list of research issues, major issues can be envisioned as follows: Multi Agent Planning: This domain may involve planning and execution monitoring for over 10,000 agents, that has different physical and informational ....

Kambhampati, S., Mali, A., and Srivastava, B., "Hybrid Planning for Partially Hierarchical Domains," Proc. of AAAI-98, Wisconsin, 1998.

....and planning is the necessity to modify the general planning algorithm. Since the worlds state could be changed by some executed operators during planning, the plan space based planning algorithm should be combined with a state space approach, like in UCP (Kambhampati and Srivastava, 1995; Kambhampati et al. 1998). An important difference to UCP is, as mentioned above, that it is not possible to backtrack over the state space plan refinements because they are already executed. User Interaction: As mentioned before, the user plays an important role in the process of information gathering. There are several ....

....abstract and concrete operators (1. and 1.A.1) This is typical for information gathering domains. So, a hierarchical planning approach seems to be usefull. WebPlan will use an extended version of CAPlan that supports hierarchical task decompositions (similar to DPOCL (Young et al. 1994) and UCP (Kambhampati et al. 1998)) combined with SNLP and dependence maintenance for interactive planning. Information Sources: The advantage to use the Internet as an information source is to have a lot of information about nearly all domains. On the other hand, the different sources in the Internet have no standardized ....

Kambhampati, S., Mali, A., and Srivastava, B. (1998). Hybrid planning for partially hierarchical domains. In Proceedings of AAAI-98.

....and Graphplan based algorithms, we close by briefly mentioning some other recent trends. 5. 1 Planning as Search Refinement search forms an elegant framework for comparing different planning algorithms and representations [51, 48] Recent results extend the theory to handle partially HTN domains [46]. McDermott showed that an emphasis on (automatically) computing an informative heuristic can make an otherwise simple planner extremely effective [81] TLPLAN uses (user provided) domain specific control information to offset a simple, forward chaining search strategy with impressive results ....

R. Kambhampati, A. Mali, and B. Srivastava. Hybrid planning for partially hierarchical domains. In Proc. 15th Nat. Conf. AI, pages 882--888, 1998.

No context found.

Subbarao Kambhampati, Amol Mali & Biplav Srivastava, Hybrid planning in partially hierarchical do- mains, Proceedings of the National Conference on Arti cial Intelligence (AAAI), 1998.

No context found.

Subbarao Kambhampati, Amol Mali and Biplav Srivastava, Hybrid planning for partially hierarchical domains, Proceedings of National Conference on Arti cial Intelligence (AAAI), Madison, pp. 882-888.

No context found.

Subbarao Kambhampati, Amol Mali & Biplav Srivastava, Hybrid planning in partially hierarchical domains, Proceedings of the National Conference on Arti cial Intelligence (AAAI), 1998.

No context found.

Subbarao Kambhampati, Amol Mali & Biplav Srivastava, Hybrid planning in partially hierarchical domains, Proceedings of the National Conference on Arti cial Intelligence (AAAI), 1998.

No context found.

Subbarao Kambhampati, Amol Mali & Biplav Srivastava, Hybrid planning in partially hierarchical domains, Proceedings of AAAI-98.

....world, as well as planning and scheduling literature. For the most part, project management in the commercial world is performed with a standard tool like Microsoft Project[38] which is primarily an event scheduling tool. There is a rich body of work in classical planning [34] 42] 24] 4] [23]. In classical planning[1] time is atomic, actions are observable and deterministic, resources are not modeled separately from other objects in the domain and there is no reward for resource optimization. The primary focus is on obtaining a sequence of actions to achieve the goal state. ....

....a complete and correct plan, RealPlan starts with a maximally parallel resource abstracted plan and add or shift actions across levels to handle resource constraints. 57 8. 6 Discussion on Plan Abstraction Abstraction and least commitment has been widely studied in the context of planning [11] [23]. In the context of making planning efficient, RealPlan can be seen as the abstraction of resources from planning to the accentuate the resource allocation problem. Specifically, only the identity of resources is abstracted into variables and the constraints (bindings) among variables are deduced ....

Kambhampati, S., Mali, A., and Srivastava, B. Hybrid planning for partially hierarchical domains In Proc. AAAI-98. July 1998.

....by increasing the length of the plan is a viable option in AI planning but not usually allowed in commercial projects due to increased costs or loss of business opportunities. 33 3. 4 Discussion on Plan Abstraction Abstraction and least commitment has been widely studied in the context of planning[9, 18], and therefore, the current implementation has focussed on a sufficient sub set to accentuate the resource allocation problem. Specifically, only the identity of resources is abstracted into variables and the constraints (bindings) among variables are deduced after an abstract plan is obtained, ....

....and plan and schedule quality can be used in deciding the specific reasoners for assembling an efficient planner. Chapter 8 Discussion and related work In this chapter, the current work is put in context of planning and scheduling literature. There is a rich body of work in classical planning[28, 36, 19, 4, 18]. In classical planning[1] time is atomic, actions are observable and deterministic, resources are not modeled separately from other objects in the domain and there is no reward for resource optimization. The primary focus is on obtaining a sequence of actions to achieve the goal state. ....

Kambhampati, S., Mali, A., and Srivastava, B. Hybrid planning for partially hierarchical domains In Proc. AAAI-98. July 1998.

No context found.

Subbarao Kambhampati, Amol Mali & Biplav Srivastava, Hybrid planning in partially hierarchical domains, Proceedings of the National Conference on Artificial Intelligence (AAAI), 1998.

No context found.

#46# R. Kambhampati, A. Mali, and B. Srivastava. Hybrid planning for partially hierarchical domains. In Proc. 15th Nat. Conf. AI, pages 882#888,

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