Journal ArticleDOI
A constraint programming formulation for planning: from plan scheduling to plan generation
TLDR
This work is based on the original model of CPT, an optimal temporal planner, and it extends the CPT’s formulation to deal with more expressive constraints, and shows that the general formulation can be used for planning and/or scheduling.Abstract:
Planning research is recently concerned with the resolution of more realistic problems as evidenced in the many works and new extensions to the Planning Domain Definition Language (PDDL) to better approximate real problems. Researchers’ works to push planning algorithms and capture more complex domains share an essential ingredient, namely the incorporation of new types of constraints. Adding constraints seems to be the way of approximating real problems: these constraints represent the duration of tasks, temporal and resource constraints, deadlines, soft constraints, etc., i.e. features that have been traditionally associated to the area of scheduling. This desired expressiveness can be achieved by augmenting the planning reasoning capabilities, at the cost of slightly deviating the planning process from its traditional implicit purpose, that is finding the causal structure of the plan. However, the resolution of complex domains with a great variety of different constraints may involve as much planning effort as scheduling effort (and perhaps the latter being more prominent in many problems). For this reason, in this paper we present a general approach to model those problems under a constraint programming formulation which allows us to represent and handle a wide range of constraints. Our work is based on the original model of $\mathsf{CPT}$ , an optimal temporal planner, and it extends the $\mathsf{CPT}$ ’s formulation to deal with more expressive constraints. We will show that our general formulation can be used for planning and/or scheduling, from scheduling a given complete plan to generating the whole plan from scratch. However, our contribution is not a new planner but a constraint programming formulation for representing highly-constrained planning + scheduling problems.read more
Citations
More filters
Journal ArticleDOI
A state of the art review of intelligent scheduling
TL;DR: A survey of intelligent scheduling systems is provided by categorizing them into five major techniques containing fuzzy logic, expert systems, machine learning, stochastic local search optimization algorithms and constraint programming.
Journal ArticleDOI
A MAS-based infrastructure for negotiation and its application to a water-right market
TL;DR: A MAS-based infrastructure for the specification of a negotiation framework that handles multiple negotiation protocols in a coherent and flexible way that may be available at any given time, to be activated and tailored on demand on demand by participating agents.
Proceedings Article
Learning Temporal Action Models via Constraint Programming.
Antonio Garrido,Sergio Jiménez +1 more
TL;DR: A solver-independent Constraint Programming formulation for learning action models in temporal planning scenarios beyond PDDL2.1 that evidences the connection between the tasks of: i) action model learning, ii) plan validation, and iii) plan synthesis.
Proceedings ArticleDOI
On the application of planning and scheduling techniques to e-learning
Antonio Garrido,Eva Onaindia +1 more
TL;DR: This paper presents an integrated planning and scheduling approach that accommodates the temporal and resource constraints of the environment to make a course applicable in a real scenario.
Automatic production planning for the construction of complex ships
Abstract: European shipyards specialize in building complex ship types including offshore vessels, yachts, dredgers, and cruise ships. One key difference between these ships and the simple cargo ships typically built in the Far East is the amount and variety of mission-related equipment required to operate the ships. Technical spaces of complex ships are numerous and densely packed. Outfitting is the shipbuilding process of installing this equipment and its supporting components (e.g. piping, ducting, and cabling). Most shipyards do not adequately plan the outfitting process. Instead, high level schedules are typically provided to outfitting subcontractors. These schedules indicate the time windows during which they must complete their installation tasks. Conflicts between the different stakeholders are addressed during weekly meetings. This outfitting planning approach is characterized by disorganization, poor communication, and a lack of transparency. As a result, the outfitting process of European shipyards is often plagued by delays, rework, and sub-optimization. A ship is constructed by first building large steel blocks, referred to as sections. Steel parts and profiles are welded together to create sections during the section building process. At the conclusion of section building, time is reserved for installing components in a section. The hull of the ship is formed by welding these sections together on a slipway or drydock. This process is referred to as erection. European shipyards mainly focus on planning the steel-related tasks of the section building and erection processes. However, their workload has shifted in recent years to become increasingly dominated by outfitting tasks. This mismatch further worsens the outfitting-related problems facing these shipyards. Automatic production planning can potentially mitigate some of the main problems facing European shipyards building complex ships. However, to maximize the effectiveness of such an approach, an integrated method must be created which considers all relevant portions of the shipbuilding process: erection, section building, and outfitting. This dissertation develops an Integrated Shipbuilding Planning Method. This method uses the characteristics of a shipyard, the geometry of a ship, and major project milestones to automatically generate an integrated erection, section building, and outfitting plan. The Integrated Shipbuilding Planning Method was not designed to replace existing shipyard planners, but instead enhance their decision-making abilities. The method aims to provide these planners with a set of high-quality production schedules that can be used as a starting point for drafting the initial plan. The foundation of Integrated Shipbuilding Planning Method is based on a mathematical model of the shipbuilding process. This model was synthesized from existing literature, expert opinion, and an analysis of the operations of a typical European shipyard. This model explicitly defines the geometric, operational, and temporal relationships that constrain the shipbuilding process. Novel techniques were developed to automatically extract several of these constraints from the data readily available in a shipyard. The mathematical model also defines the objectives used to measure the quality of a production schedule. A combination of multi-objective genetic algorithms and custom designed heuristics were used to solve the proposed mathematical model. This solution approach tailored historically successful optimization techniques to the specific problem structure of scheduling shipbuilding tasks. Although the developed solution approach does not guarantee that the optimal solution will be found, it allows for sufficiently high-quality solutions to be discovered in reasonable computational times. The Integrated Shipbuilding Planning Method was evaluated with a test case of a pipelaying ship recently delivered from a Dutch shipyard. This method created a variety of high-quality production plans of both the erection and section building processes in a reasonable computational time. The automatically generated production schedules significantly outperformed those manually generated by the shipyard planners. Especially large gains were seen with respect to the evenness of the outfitting workload and the time available to install components on the slipway. Furthermore, the negligible run time allows planners to quickly make adjustments and test different scenarios. The input data required for creating the section building and erection schedules matches the information that shipyard planners have access to at the start of a new project. Not only was the Integrated Shipbuilding Planning Method able to optimize the planning of the erection and section building independently, it was also shown to be capable of concurrently optimizing the planning of both processes. Implementing the Integrated Shipbuilding Planning Method in a shipyard for automatically scheduling the section building and erection processes should be relatively straightforward. This method works with the same data (both input and output) as the shipyard planners drafting the initial production schedules. A shipyard would still need to adapt the method to their own process by incorporating their own production data; modifying the constraints and objective to match their production process; tuning the parameters of the solution technique; and implementing the result in the work flow of their planners. However, the global approach and algorithms underlying the solution technique are directly applicable. A detailed outfitting schedule was also created for the test case ship using the Integrated Shipbuilding Planning Method. Although a high-quality solution was found, the required computational time was somewhat extensive due to the large problem size and complex nature of the relationships constraining the installation of outfitting components. The detailed outfitting schedule was used to determine the influence of the outfitting process on erection and section building. To generate the detailed outfitting schedule, a high level of geometric detail was required because such a schedule is defined on the component level. Such detailed geometry, however, is generally not fully available prior to the onset of outfitting due to the concurrent nature of the detailed engineering and production processes of modern European shipyards. The full implementation of the Integrated Shipbuilding Planning Method for automatically generating detailed outfitting schedules is currently limited by the extensive computational requirements and the timely availability of detailed geometric data. The Integrated Shipbuilding Planning Method was also used to examine two production scenarios to demonstrate its applicability in making strategic decisions. The method was first used to evaluate the performance of three different block building strategies in relation to the erection and section building processes. A recommendation was given for the best strategies assuming the shipyard prioritized having a level resource demand. The effect of the implementation of multi-skilled workers on the outfitting process was also examined. This scenario determined the effect of six different types of multi-skilled mounting teams on the total number of mounting teams required to build the test case ship. In both cases, the scenario analyses provided additional, useful information which could aid a shipyard in making strategic decisions. Because strategic decisions are generally based on historical data, the timely availability of detailed geometric data should not hinder the applicability of the Integrated Shipbuilding Planning Method for supporting such decisions. The Integrated Shipbuilding Planning Method is novel for several reasons. First, this method is the only automatic planning method developed for shipbuilding that fully incorporates the outfitting process. This method is also the first example of a scheduling methodology that concurrently plans the erection and section building tasks of a shipbuilding project. Furthermore, this approach demonstrates the feasibility of using a priority-based heuristic function in a multi-objective genetic algorithm to effectively schedule a large set of production tasks. Lastly, the production scenarios examined using the Integrated Shipbuilding Planning Method prove that it is possible for a shipyard to use optimization techniques to support strategic planning decisions.
References
More filters
Journal ArticleDOI
The FF planning system: fast plan generation through heuristic search
Jörg Hoffmann,Bernhard Nebel +1 more
TL;DR: A novel search strategy is introduced that combines hill-climbing with systematic search, and it is shown how other powerful heuristic information can be extracted and used to prune the search space.
Journal ArticleDOI
Temporal constraint networks
TL;DR: It is shown that the STP, which subsumes the major part of Vilain and Kautz's point algebra, can be solved in polynomial time and the applicability of path consistency algorithms as preprocessing of temporal problems is studied, to demonstrate their termination and bound their complexities.
Proceedings Article
Fast planning through planning graph analysis
Avrim Blum,Merrick L. Furst +1 more
TL;DR: A new approach to planning in STRIPS-like domains based on constructing and analyzing a compact structure the authors call a Planning Graph is introduced, and a new planner, Graphplan, is described that uses this paradigm.
Book
STRIPS: a new approach to the application of theorem proving to problem solving
Richard Fikes,Nils J. Nilsson +1 more
TL;DR: In this article, the authors describe a problem solver called STRIPS that attempts to find a sequence of operators in a spcce of world models to transform a given initial world model into a model in which a given goal formula can be proven to be true.
Book
Automated Planning, Theory And Practice
TL;DR: This chapter discusses Classical Planning and its Applications, as well as Neoclassical and Neo-Classical Techniques, and discusses search procedures and Computational Complexity.