Smith, D. R. Synthesis of high-performance transportation schedulers. Tech. Rep. KES.U.95.6, Kestrel Institute, March 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....and Scheduling Technology Over the past several years, Rome Laboratory has been evaluating the use of advanced AI planning and scheduling technology in large, real world applications. This technology has been successfully applied to military transportation and airlift scheduling in the KTS system [19, 20] and ITAS system [2] and nuclear power plant outage management in the ROMAN system [1, 8, 9] In this section, we succintly describe this advanced AI planning and scheduling technology and the different application domains. A central aspect of this technology is the use of a semi automatic ....

....rest of the partial schedule. As previously mentioned, several applications have been developed to demonstrate the effectiveness of these techniques in the domains of military transportation and airlift scheduling and nuclear power plant outage management. KTS (Kestrel Transportation Scheduler) [19, 20] was developed mainly as a technology demonstration and used a realistic model of the military transportation scheduling domain for the US Transportation Command. KTS schedules 15,460 individual movement requirements in 71 cpu seconds on a Sparc 2. Furthermore, the resulting schedules produced by ....

Douglas R. Smith, Eduardo Parra, and Stephen Westfold. Synthesis of High Performance Transportation Schedulers. Technical Report KES.U.95.1, Kestrel Institute, 1995.

....given problem specification. KIDS uses a form of deductive inference called directed inference to reason about the problem specification in order to automatically apply the various transformations [5] KIDS has been successfully used in the derivation of high performance transportation schedulers [6,7,11]. These applications have shown that advanced planning and scheduling technology is beneficial in complex and realistic problem domains. However, the development of a KIDS domain theory, the selection and implementation of a search strategy, and knowing what sequence of transformations to apply to ....

....features that ROMAN provides. Roman generates schedules incorporating very complex constraints as in the safety constraints for AC Power. The constraint model used in ROMAN is more general than the models used in previous scheduling applications using KIDS technology such as KTS and ITAS [6,7,11] particularly with regard to the way Maximum on Ground (MOG) port constraints are handled. ROMAN provides increased robustness in terms of schedules that are feasible over time intervals rather than a single time point as start times. Thus, ROMAN s solution provides a family of schedules rather ....

Smith, Douglas R., Parra, Eduardo, and Westfold, Stephen. Synthesis of High Performance Transportation Schedulers. Technical Report KES.U.95.1, Kestrel Institute, 1995.

....problem specification. Finite differencing is another important transformation provided by KIDS. KIDS uses a form of deductive inference called directed inference to reason about the problem specification in order to automatically apply tactics, derive filters and perform constraint propagation[Smith et al. 95] KIDS has been used to derive a very fast transportation scheduler for the US Transportation Command, KTS (Kestrel Transportation Scheduler) Smith Parra 93] A typical transportation problem with 10; 000 movement requirements takes the derived scheduler 1 to 3 minutes to solve, compared with ....

....feasibility estimator (JFAST) and 36 hours for deployed schedulers (FLOGEN, ADANS) The computed schedules use relatively few resources and satisfy all specified constraints. The speed of this scheduler was due to the synthesis of strong constraint checking and constraint propagation code [Smith et al. 95] In this paper we show how this approach can be extended to tackle a much richer real world scheduling task involving complex state variables and time windows. 3 Planning and Scheduling of Nuclear Power Plant Outages The planning and scheduling of the operations involved in the outages of ....

[Article contains additional citation context not shown here]

Doug Smith, Eduardo Parra, and Stephen Westfold. Synthesis of High Performance Transportation Schedulers. Technical Report Tech. Rep. KES.U.95.1, Kestrel Institute, 1995. 77

....the propositions and terms in subcontexts to propositions and terms in an outer context. Here is an example. Assume that a route planner, like the route optimization program of the TRAINS project (Allen et al. 1995) and a supply planner, like the transportation scheduler developed at Kestrel (Smith, Parra, Westfold 1995), have been developed independently by different groups. Given a source and a destination, the route planner will find the best route between these places. It however, has no notion of which supplies need to be transported and no notion of time. The supply planner keeps track of the supplies of ....

Smith, D. R.; Parra, E. A.; and Westfold, S. J. 1995. Synthesis of High-Performance Transportation Schedulers. Technical Report KES.U.95.1, Kestrel Institute. http://kestrel.edu/www/publications.html.

....that relate the propositions and terms in subcontexts to propositions and terms in an outer context. Here is an example. Assume that a route planner, like the route optimization program of the TRAINS project [3] and a supply planner, like the transportation scheduler developed at Kestrel [59], have been developed independently by different groups. Given a source and a destination, the route planner will find the best route between these places. It however, has no notion of which supplies need to be transported and no notion of time. The supply planner keeps track of the supplies of ....

D. R. Smith, E. A. Parra, and S. J. Westfold. Synthesis of High-Performance Transportation Schedulers. Technical Report KES.U.95.1, Kestrel Institute, 1995. http://kestrel.edu/www/publications.html.

....term KIDS approach to denote the concepts that have been implemented in the system KIDS 1 . The KIDS approach has been applied to a number of case studies at Kestrel Institute. In particular, it has been used in the design of a transportation scheduling algorithm with impressive performance [15, 16]. We wished to find out if we were able to use this method based on the available publi 1 We did not use the implemented system KIDS in the case study. cations and produce satisfactory results with reasonable effort. A second goal of this work has been to study how a knowledge based approach ....

....who wish to learn and use them. However, we did not use the KIDS system because we wanted to have full control over the design process and adapt it to our needs if necessary. Transportation scheduling. Our case study relates to the research on design of transportation schedulers at Kestrel [15, 16]. They study schedulers that assign trips to resources like planes, ships, and trucks to meet movement requirements. In this setting, trips fully occupy resources for an interval of time, i.e. the load of a resource cannot be extended during a trip. Furthermore, a trip changes the availability of ....

[Article contains additional citation context not shown here]

D. R. Smith, E. A. Parra, and S. J. Westfold. Synthesis of high-performance transportation schedulers. Technical Report KES.U.95.6, Kestrel Institute, 1995.

....of the approach were used in the case study, and could modify the approach where necessary. The KIDS system has been applied to a number of case studies at Kestrel Institute. In particular, it has been used in the design of a transportation scheduling algorithm with impressive performance [18] [19]. We wished to find out if we were able to use this method based on the available publications and produce satisfactory results with reasonable effort. A second goal of this work has been to study how a knowledge based approach can be integrated into the overall software engineering 2 process. ....

....kernel, so new ones can be integrated into the system in a routine way. The system Specware [20] under development at Kestrel also seems to allow for a modularized and easily extendible knowledge base. Our case study relates to the research on design of transportation schedulers at Kestrel [18] [19]. They study schedulers that assign trips to resources like planes, ships, and trucks to meet movement requirements. In this setting, trips fully occupy resources for an interval of time, i.e. the load of a resource cannot be extended during a trip. Furthermore, a trip changes the availability of ....

[Article contains additional citation context not shown here]

D. R. Smith, E. A. Parra, and S. J. Westfold. Synthesis of high-performance transportation schedulers. Technical Report KES.U.95.6, Kestrel Institute, 1995.

....to accomplish the ideal optimization. Later we will see this idea arising in other approaches that we will discuss: GenVoca [Batory, Chen, et al. 1997b] Anticipatory Optimization [Biggerstaff 1997, 1998] Aspect Oriented Programming [Kiczales et al. 1997] and Kids [Smith 1990, 1991; Smith et al. 1996]. Separate domains and small grain transformations of the Draco variety are not without problems. 3.1.7 Global Dependencies The problem with small grain transformations used in a system that operates by pure forward refinement (i.e. by a local substitution paradigm) is that such a system does ....

....such domains. Another difficulty arises because of the search space induced by the deep inference chains. This is likely to lead to generational inefficiency in comparison to Draco or GenVoca style systems. Nevertheless, this is promising work that is beginning to provide practical contributions [Smith et al. 1996]. Systems in this class had their roots in the early work of Green [1969] and Waldinger [1969] who developed methods for developing constructive proofs of the existence of a program that met a given Input Output specification. The desired program was derived as a side effect of the proof process. ....

Smith, D. R., E. A. Parra, S. J. Westfold (1996), "Synthesis of High-Performance Transportation Schedulers," In Advanced Planning Technology, Ed. A. Tate, AAAI Press, Menlo Park, CA. (See also http://www.kestrel.edu/HTML/publications.html).

.... [Smi 90] Many case studies have proven its feasibility and demonstrated how much more abstract and user oriented developments could be achieved than using usual post verification approaches (fundamental for systems like PVS [OSR 93] One recent case study is [KW 95] and a prominent one is [SPW 95] where a strategic transportation scheduling algorithm is developed which is 200 times faster than the ones in practical use today. Unfortunately, implementations of transformation systems tend to be complicated and insecure. The correctness issue of transformation rules is usually not treated at ....

....by the auxiliary function Fgs: if z is not directly extractable from r this search space is split and its subspaces are searched. The global search theory described above is relatively simple. More refined ones incorporate filters to prune search spaces. The most elaborate one stated in [SPW 95] uses a refinement relation on search spaces and cutting constraints to profoundly exploit the problem domain and synthesise highly efficient search algorithms. 3.2 Formalisation in Isabelle HOL How do we know that a particular application of Global Search is correct, i.e. how can we be sure ....

[Article contains additional citation context not shown here]

Smith, D. R., Parra, E. A., Westfold, S. J.: Synthesis of High-Performance Transportation Schedulers, Technical Report, Kestrel Institute, Palo Alto, 1995.

No context found.

Smith, D. R. Synthesis of high-performance transportation schedulers. Tech. Rep. KES.U.95.6, Kestrel Institute, March 1995.

....scheduling model, and furthermore, during execution the schedule will need to be periodically revised to adapt to unanticipated circumstances. As part of the ARPA Rome Laboratory Planning and Scheduling Initiative, we have focused on the transformational development of transportation schedulers [12, 14]. Our approach involves several stages. The first step is to develop a formal model of the transportation scheduling domain, called a domain theory. Second, the constraints, objectives, and preferences of a particular scheduling problem are stated within a domain theory as a problem ....

....[16, 11, 13] 3 Synthesizing Schedulers In this section we informally describe some of the algorithmic knowledge that is used to synthesize schedulers. Formal presentation of this knowledge and the deductive machinery necessary to apply it to concrete problem specifications may be found in [7, 14]. The general idea underlying the algorithm design tactics in KIDS is to represent abstract knowledge about a class of algorithms (such as divide and conquer or dynamic programming) as a theory, called an algorithm theory [9] Models of an algorithm theory correspond to instances for a particular ....

[Article contains additional citation context not shown here]

Smith, D. R., Parra, E. A., and Westfold, S. J. Synthesis of high-performance transportation schedulers. Tech. Rep. KES.U.95.6, Kestrel Institute, March 1995.

.... 10; 000 movement requirements takes the derived scheduler 1 to 3 minutes to solve, compared with 2:5 hours for a deployed feasibility estimator (JFAST) and 36 hours for deployed schedulers (FLOGEN, ADANS) The computed schedules use relatively few resources and satisfy all specified constraints [11]. In this paper we show how previous applications of KIDS to scheduling problems can be extended to tackle a much richer real world scheduling task with planninglike features involving complex safety constraints and resource constraints in the presence of time windows on each activity, a novel ....

....A ] A C [ A C F ] A C G F [ A C F G ] spliitting splitting cutting splitting cutting splitting cutting est lst A, B,C, activities Figure 5. Global search theory for the Outage Problem 4.2. Global Search Theory Global search [12, 11] is a backtrack algorithm, a refinement of generate and test. The tactic is implemented by finding a space containing all the solutions to the problem that can be divided into nested subspaces. The global search algorithm starts with an initial set that contains all the solutions to the given ....

[Article contains additional citation context not shown here]

D. Smith, E. Parra, and S. Westfold. Synthesis of High Performance Transportation Schedulers. Technical Report Tech. Rep. KES.U.95.1, Kestrel Institute, 1995.

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