Automated Planning

In most real-world reasoning problems, planning and scheduling phases are loosely coupled. For example, in project planning, the user comes up with a task list and schedules it with a scheduling tool like Microsoft Project. One can view automated planning in a similar way in which there is an action selection phase where actions are selected and ordered to reach the desired goals, and a resource allocation phase where enough resources are assigned to ensure the successful execution of the chosen actions. On the other hand, most existing automated planners studied in Artificial Intelligence do not exploit this loose-coupling and perform both action selection and resource assignment employing the same algorithm. The current work shows that the above strategy severely curtails the scale-up potential of existing state of the art planners which can be overcome by leveraging the loose coupling. Specifically, a novel planning framework called RealPlan is developed in which resource allocation is de-coupled from planning and is handled in a separate scheduling phase. The scheduling problem with discrete resources is represented as a Constraint Satisfaction Problem (CSP) problem, and the planner and scheduler interact either in a master-slave manner or in a peer-peer relationship. In the former, the scheduler simply tries to assign resources to the abstract causal plan passed to it by the planner and returns success. In the latter, a more sophisticated "multi-module dependency directed backtracking" approach is used where the failure explanation in the scheduler is translated back to the planner and serves as a nogood to direct planner search. RealPlan not only preserves both the correctness as well as the quality (measured in length) of the plan but also improves efficiency. Moreover, the failure-driven learning of constraints can serve as an elegant and effective approach for integrating planning and scheduling systems. Beyond the * Corresponding author.

By relaxing the hard-goal constraints from classical planning and associating them with reward values, over-subscription planning allows users to concentrate on presenting what they want and leaves the task of deciding the best goals to achieve to the planner. In this paper, we extend the over-subscription planning problem and its limited goal specification to allow numeric goals with continuous utility values and goals with mixed hard and soft constraints. Together they considerably extend the modeling power of goal ...

Due to the recent development of deep water fields, the mechanical loads in flexible risers have been significantly increased and very high strength wires are often needed (SMUTS ≥ 1200 MPa). Moreover, the development of deep water fields containing H2S leads to consider some H2S content during the design phase. Steel wires are not in direct contact with the conveyed fluid, but are located in the annular space which is a specific medium for corrosion less severe than the bore. The new very high strength steel presented in this paper is compatible with annulus environment containing H2S and is very beneficial for designing deep water risers.

Complete planning and scheduling of all spacecraft operations is a challenging area with the remote agent experiment at Deep Space 1 being a pioneering system. Still the complete approach is rare in practice. For example, in the Mars Express (MEX) mission, planning and scheduling techniques are used to solve some subproblems namely scheduling command upload and data download. In this paper we describe an approach to generate a complete schedule of the spacecraft that includes planning and scheduling of science, command uplink, data downlink, maintenance, and pointing operations. The proposed solving approach was designed to plan operations on the Mars Express (MEX) mission and it was motivated by the MEX challenge at the Fourth International Competition on Knowledge Engineering for Planning and Scheduling. The method is based on incremental addition of operations to a partial schedule and modifying the time allocation of already scheduled operations to fit the newly added operation....

A dynamic and complex process environment forces the automated execution and monitoring of processes to face a lot of hard problems. This paper shows how the execution of agile workflows can be assisted by AI planning techniques. In contrast to previous approaches, this method allows for an online adaption of simple process models to changes in the process environment. It can handle a manifold of unexpected events and guarantees the soundness and completeness of the adapted workflow.

Purpose: Inconsistencies in contouring target structures can undermine the precision of conformal radiation therapy (RT) planning and compromise the validity of clinical trial results. This study evaluated the impact of guidelines on consistency in target volume contouring for partial breast RT planning. Methods and Materials: Guidelines for target volume definition for partial breast radiation therapy (PBRT) planning were developed by members of the steering committee for a pilot trial of PBRT using conformal external beam planning. In phase 1, delineation of the breast seroma in 5 early-stage breast cancer patients was independently performed by a "trained" cohort of four radiation oncologists who were provided with these guidelines and an "untrained" cohort of four radiation oncologists who contoured without guidelines. Using automated planning software, the seroma target volume (STV) was expanded into a clinical target volume (CTV) and planning target volume (PTV) for each oncologist. Means and standard deviations were calculated, and two-tailed t tests were used to assess differences between the "trained" and "untrained" cohorts. In phase 2, all eight radiation oncologists were provided with the same contouring guidelines, and were asked to delineate the seroma in five new cases. Data were again analyzed to evaluate consistency between the two cohorts. Results: The "untrained" cohort contoured larger seroma volumes and had larger CTVs and PTVs compared with the "trained" cohort in three of five cases. When seroma contouring was performed after review of contouring guidelines, the differences in the STVs, CTVs, and PTVs were no longer statistically significant. Conclusion: Guidelines can improve consistency among radiation oncologists performing target volume delineation for PBRT planning.

Large-scale space structures have taken center stage as the future
of space exploration, paving the way for ambitious endeavors like
sprawling solar power stations, intricate telescopes, and giant space
stations. This program lies not in manual assembly, but in empowering space robots to autonomously build these structures through
plans provided by dedicated on-orbit assembly planning algorithms.
In this paper we describe an approach to large-scale space structures assembly frameworks, while keeping in mind the many benefits associated to the usage of autonomous crawling mobile robots.
We thus describe an assembly approach and model on which
we are developing our research. Initially, all the building elements
(beams and nodes) are stored in the payload fairing. These are
deployed bit by bit building the structure on which the robots themselves will evolve until reaching the final configuration. We represent the assembling problem as an automated planning instance,
where several structural constraints dictate the actions available
for execution. The automated planner algorithm provides the steps
required to achieve the final, deployed structure.
We implemented a first version of the planner that provides an
assembly plan, i.e. a sequence of actions to move the elements from
the payload fairing to the deployed structure.
This model helps to illustrate the feasibility of the approach and
the benefits of having AI tools for autonomous assembling robots
in space.

The introduction of PDDL+ allowed more accurate representations of complex real-world problems of interest to the scientific community. However, PDDL+ problems are notoriously challenging to planners, requiring more advanced heuristics. We introduce the Temporal Pattern Database (TPDB), a new domain-independent heuristic technique designed for PDDL+ domains with mixed discrete/continuous behaviour, non-linear system dynamics, processes, and events. The pattern in the TPDB is obtained through an abstraction based on time and state discretisation. Our approach combines constraint relaxation and abstraction techniques, and uses solutions to the relaxed problem, as a guide to solving the concrete problem with a discretisation fine enough to satisfy the continuous model's constraints.

The introduction of PDDL+ allowed more accurate representations of complex real-world problems of interest to the scientific community. However, PDDL+ problems are notoriously challenging to planners, requiring more advanced heuristics. We introduce the Temporal Pattern Database (TPDB), a new domain-independent heuristic technique designed for PDDL+ domains with mixed discrete/continuous behaviour, non-linear system dynamics, processes, and events. The pattern in the TPDB is obtained through an abstraction based on time and state discretisation. Our approach combines constraint relaxation and abstraction techniques, and uses solutions to the relaxed problem, as a guide to solving the concrete problem with a discretisation fine enough to satisfy the continuous model’s constraints.

Automated planning is a prominent Artificial Intelligence (AI) challenge that has been extensively studied for decades, which has led to the development of powerful domain-independent planning systems. The performance of domain-independent planning systems are strongly affected by the structure of the search space, that is dependent on the application domain and on its encoding. This paper proposes and investigates a novel way of combining machine learning and heuristic search to improve domain-independent planning. On the learning side, we use learning to predict the plan cost of a good solution for a given instance. On the planning side, we propose a bound-sensitive heuristic function that exploits such a prediction in a state-space planner. Our function combines the input prediction (derived inductively) with some pieces of information gathered during search (derived deductively). As the prediction can sometimes be grossly inaccurate, the function also provides means to recognise when the provided information is actually misguiding the search. Our experimental analysis demonstrates the usefulness of the proposed approach in a standard heuristic best-first search schema.

Recently, a new decomposition based relaxation has been proposed for numeric planning problems. Roughly, this relaxation is grounded on the identification of regression-based necessary conditions for the satisfaction of sets of numeric subgoals. So far, it has been used to define novel heuristics that are able to provide great guidance in problems exhibiting a pronounced numeric structure. This paper investigates how to further exploit this relaxation; it does so by introducing the notion of the multi-repetition relaxed plan. The multi-repetition plan annotates actions with the number of times such actions need to be executed. We use this structure for different purposes: extraction of a concrete relaxed plan based heuristic, definition of subgoaling based helpful actions, and definition of what we call up-to-jumping actions. Up-to-jumping actions allow us to deeply leverage from the metric structure of the problem and devise an informed search strategy that can collapse several dec...

The main objective of my research activity is the development and the experimental analysis of efficient techniques of domain independent planning, which proposes to identify a partially ordered set of actions whose execution allows one or more agents to reach their goals starting from a specific initial situation. My research has been focused on generic algorithms that can be applied to a variety of situations in which the evolution of the external world makes it necessary to revise a plan which has been previously formulated or the generation of a new plan. In such a context I have developed the systems LPG, LPG-td and ADJ together with Prof. Alfonso Gerevini and other members of the AI group of the University of Brescia. In particular LPG is at present one of the best domain independent planners that exist in terms of execution times, quality of the plans produced and expressivity of manageable problems. LPG took part in the third International Planning Competition IPC3 obtaining...

Despite recent progress in planning, many complex domains and even larger problems in simple domains remain hard and challenging. One way to achieve further improvement is to utilise knowledge acquired for the planner from the domain. Macro-actions are a promising means by which to convey such knowledge. A macro-action, or macro in short, is a group of actions selected for application as a single choice. Most existing works on macros utilise knowledge explicitly specific to the planners and the domains. But ...

Despite much research into Artificial Intelligence (AI) and Intelligent Agents (IA) over the past few years little 'intelligent' behaviour is displayed by modern computer systems. AI planning and intelligent systems are failing to be generally used. The intellectual abilities such as learning, analysing, problem solving, planning and abstracting are features that we do not entirely associate with today's computer systems. In AI we recognise that although machines are beginning to overtake the human brain in terms of sheer processing power and perhaps storage capacity, they still cannot approach the level of human intelligence in terms of general purpose cognitive capability. Part of the reason for this is that software technology that supports general intelligent processes is difficult to use, understand and maintain. With the advent of the Internet, the Web as we know it is an enormous collection of information stored in a variety of formats and held loosely together by hyperlinks. This is adequate for humans who are able to understand natural languages, but it is of no use if we want machines to be able to process as well as understand the information. To deal with this issue, the next Web generation promise to deliver Semantic Web Services; services that are self-described and amenable to automated discovery, composition and choreography. They can be utilised by application or other services without human assistant or protocols; where information is encoded according to well defined vocabularies, often known as Ontologies. In this research, we aim to create the infrastructure to enable intelligent software to be recast as semantic web services, and lead the way to "Service Oriented Intelligence"distributed software intelligence using the Internet.

One way to improve the robustness and flexibility of robot performance is to let the robot learn from its experiences. In this paper, we describe the architecture and knowledge-representation framework for a service robot being developed in the EU project RACE, and present examples illustrating how learning from experiences will be achieved. As a unique innovative feature, the framework combines memory records of low-level robot activities with ontology-based high-level semantic descriptions.

Artykuł wprowadza w problematykę związaną z modelowaniem egzoszkieletu kończyn dolnych z napędem hydraulicznym. Na początku przedstawiono kinematyczny model egzoszkieletu. Wyznaczono położenie każdego z charakterystycznych punktów w przyjętym układzie współrzędnych z wykorzystaniem macierzy kosinusów. Następnie określono zmianę długości każdego z siłowników hydraulicznych podczas chodu i biegu. Drugi etap badań polegał na zamodelowaniu poszczególnych części egzoszkieletu w programie CAD w celu wyznaczenia momentów bezwładności każdego z elementów. Bazując na zaproponowanym modelu egzoszkieletu opracowano równania dynamiki i przeprowadzono symulację zmiany pozycji z przysiadu do pozycji stojącej w czasie 1 s. Uzyskane wyniki zaprezentowano na wykresach. We wnioskach końcowych nakreślono kierunki dalszych badań. WSTĘP Wymagania stawiane obecnie projektowanym egzoszkieletem są coraz większe i wraz z ich wzrostem rośnie też zakres przeprowadzanych analiz. Do ich otrzymania wykorzystywane są systemy komputerowe, które zastępują człowieka w wielu czasochłonnych operacjach obliczeniowych. Jako napędy egzoszkieletów można wykorzystać systemy pneumatyczne, hydrauliczne i elektryczne. W głównej mierze jest to uzależnione od przeznaczenia urządzenia. Napędy hydrauliczne charakteryzują się dużym współczynnikiem mocy do masy urządzeń. Do ich wad należy zaliczyć możliwość wycieków, oraz konieczność zastosowania odpowiednich pomp w celu wytworzenia wysokiego ciśnienia. Ciśnienie pracy czynnika roboczego w układach hydraulicznych wynosi od 20 do 50 MPa. W artykule przedstawiono kinematykę i dynamikę modelu egzoszkieletu o napędzie hydraulicznym.

Authoring tools for building Intelligent Educational Systems must provide support to ensure flexibility, adaptability of content to the user profile, reusability and sharing of learning objects. These facilities are essential to develop automated decision processes for providing course compositions tailored to the specific characteristics of each individual learner. We present a LOM-compliant learning approach that uses an automated planning process to create personalised learning courses while giving special attention to the development of reusable learning objects.

K-12 education standards in the U.S. require all students to read complex texts across many subject areas. The Language Muse TM Activity Palette is a web-based language-instruction application that uses NLP algorithms and lexical resources to automatically generate language activities and support English language learners' content comprehension and language skills development. The system's online platform for activity generation, scoring, and feedback is scalable for MOOCs, as well as for other online learning settings.

This paper presents three approaches to generate learning designs using existing domain-independent planners. All of the approaches compile a course defined in a standard e-learning language into a planning domain, and a file containing student's learning information into a planning problem. The learning designs are automatically generated from the plans that solve the problems. The approaches differ in the kind of planning domain generated, thus increasing the possibilities of using existing planners: i) hierarchical, ii) including PDDL actions with conditional effects, and iii) including PDDL durative actions. We also analyse the pros and cons on the knowledge engineering procedures used in each approach.

In recent years the concept of sequential portfolio has become an important topic to improve the performance of modern problem solvers, such as SAT engines or planners. The PbP planner and more recently Fast Downward Stone Soup are successful approaches in Automated Planning that follow this trend. However, neither a theoretical analysis nor formal definitions about sequential portfolios have been described. In this paper, we focus on studying how to evaluate the performance of planners defining a baseline for a set of problems. We present a general method based on Mixed-Integer Programming to define the baseline for a training data set. In addition to prior work, we also introduce a short empirical analysis of the utility of training problems to configure sequential portfolios.

To evaluate the feasibility of an automatic, whole-heart segmentation algorithm for measuring global heart function from gated, whole-chest MDCT images. Material and methods: 15 patients with suspicion of PE underwent whole-chest contrast-enhanced MDCT with retrospective ECG synchronization. Two observers computed right and left ventricular functional indices using a semi-manual and an automatic whole-heart segmentation algorithm. The two techniques were compared using Bland-Altman analysis and paired Student's t-test. Measurement reproducibility was calculated using intraclass correlation coefficient. Results: Ventricular analysis with automatic segmentation was successful in 13/15 (86%) and in 15/15 (100%) patients for the right ventricle and left ventricle, respectively. Reproducibility of measurements for both ventricles was perfect (ICC: 1.00) and very good for automatic and semi-manual measurements, respectively. Ventricular volumes and functional indices except right ventricular ejection fraction obtained from the automatic method were significantly higher for the RV compared to the semi-manual methods. Conclusions: The automatic, whole-heart segmentation algorithm enabled highly reproducible global heart function to be rapidly obtained in patients undergoing gated whole-chest MDCT for assessment of acute chest pain with suspicion of pulmonary embolism.

One way to improve the robustness and flexibility of robot performance is to let the robot learn from its experiences. In this paper, we describe the architecture and knowledge-representation framework for a service robot being developed in the EU project RACE, and present examples illustrating how learning from experiences will be achieved. As a unique innovative feature, the framework combines memory records of low-level robot activities with ontology-based high-level semantic descriptions.

This paper will describe the role of Human-Computer Interaction (HCI) design in the development and deployment of mission systems. We will make the argument that more broad application of HCI methods and practice will be necessary to complete the ambitious objectives of future missions. In particular, the introduction of increased autonomy, reduction of ops personnel, increasingly tactical nature of surface operations, and more distributed mission ground operations present unique challenges that must be met with sound HCI-driven solutions. We will describe a case study from where solid underlying technology was hindered by lack of attention to HCI issues, and describe how this is being remedied moving forward. Future missions will require planning processes to be much more efficient (orders of magnitude) than what was achieved for the Mars Exploration Rovers mission (2003). The case study will focus on the critical shift from the MAPGEN software used for tactical activity planning on MER to SPIFe, now base-lined for the Phoenix 2007 and MSL 2009 missions. MAPGEN added automated planning functionality (Europa) to the APGEN manual planning tool. SPIFe provides a next generation interface designed specifically to take advantage of an automated planner. The design of SPIFe is based on hundreds of hours of observation and operations experience on MER. This paper will examine how key HCI lessons learned are being fed forward into this new set of planning tools, SPIFe and Europa 2. Planning at NASA has largely been done "by hand," that is, without the use of tools to reason about complex constraint interactions, flight rule violations, etc. This has several shortcomings, but chief among them are time required to complete the task in this way and the potential for error. Even when state of the art tools are used, there are often some fundamental process limitations. Especially when under time

We discuss an application of rough set tools for modeling networks of classifiers induced from data and ontology of concepts delivered by experts. Such networks allow us to develop strategies for automated planning of a treatment of infants with respiratory illness. We report results of experiments with the networks of classifiers used in automated planning of the treatment of newborn infants with respiratory failure. The reported experiments were performed on medical data obtained from the Neonatal Intensive Care Unit in the

SAC-D/Aquarius is a joint satellite space mission with CONAE and NASA as main partners. For this mission, a distributed planning approach has been operational since the spacecraft launch on June 10, 2011. In this paper we describe the representation used by the distributed planning system that runs at the Mission Operations Center. We describe how this representation is exploited for automatic generation of all activities related with periodic orbital events and for an automatic verification of a successful plan execution. The former takes advantage of an abstraction layer provided by the planning system and the latter is based on the model inside the planning system that connects all activities considered when planning with their detailed implementation.

Computer-controlled treatment machines and the delivery of conformal radiotherapy treatments using computer-controlled and / or automated delivery techniques have become an important new area of clinical and research activity in recent years. The historical experience with simple multileaf collimators and control systems used for dynamic conformal therapy has been expanded with the development of a number of different methods used to deliver modern conformal therapy treatments. Various approaches to these automated and semi-automated treatments are described. Safety considerations are a critical part of the automated delivery process, and are also briefly reviewed. Early experience with these techniques at a number of centers is briefly described.

In many practical domains, planning systems are required to reason about durative actions. A common assumption in the literature is that the executor is allowed to decide the duration of each action. However, this assumption may be too restrictive for applications. In this paper, we tackle the problem of temporal planning with uncontrollable action durations. We show how to generate robust plans, that guarantee goal achievement despite the uncontrollability of the actual duration of the actions. We extend the state-space temporal planning framework, integrating recent techniques for solving temporal problems under uncertainty. We discuss different ways of lifting the total order plans generated by the heuristic search to partial order plans, showing (in)completeness results for each of them. We implemented our approach on top of COLIN, a stateof-the-art planner. An experimental evaluation over several benchmark problems shows the practical feasibility of the proposed approach.

The use of plastic pipelines allows to increase the effi ciency of pipeline systems of a ship due to their extraordinary corrosion resistance, a decrease of maintenance frequency, a reduction of total operational costs, and opens a new direction in the use of plastics. The weight of plastic pipelines is considerably less than that of metal ones, which allows to decrease the weight of the vessel by several tons and to increase, due to this, its maneuverability, as well as to decrease fuel consumption, ensure easier passing of all the necessary weight regulations, and optimize vessel operation. The process of installation work performance provides for a possibility of a pre-assembly of structures, the execution of welded, glued, mechanical joints with the help of a minimum set of tools in a confi ned space, and decreases by 50% the time of pipeline laying on the ship. Moreover, plastic pipe manufacturing creates fi ve times less carbon dioxide emissions, which makes it a signifi cant environmentally-friendly alternative to metal pipes. Therefore, in the cases where it is technically possible, plastic pipelines can be used as a direct substitution of metal pipelines. Sažetak Uporaba plastičnih cjevovoda omogućuje povećanje učinkovitosti brodskih cjevovodnih sustava zbog njihove izvanredne otpornosti na koroziju, smanjenja učestalosti održavanja, smanjenja ukupnih operativnih troškova, te otvara novi smjer u uporabi plastike. Težina plastičnih cjevovoda znatno je manja od metalnih, što omogućuje smanjenje težine plovila za nekoliko tona, a time i bolju manevarsku sposobnost, kao i smanjenje potrošnje goriva. Olakšava provedbu uredbi o nužnoj težini i optimizira rad plovila. Proces izvođenja instalacijskih radova pruža mogućnost predmontaže konstrukcija, izvođenja zavarenih, lijepljenih, mehaničkih spojeva uz pomoć minimalnog seta alata u skučenom prostoru, te za 50 % skraćuje vrijeme postavljanja cjevovoda na brodu. Proizvodnja plastičnih cijevi stvara pet puta manje emisija ugljičnog dioksida, što je čini značajnom ekološki prihvatljivom alternativom metalnim cijevima. Zbog toga se u slučajevima kada je to tehnički moguće, plastični cjevovodi mogu koristiti kao izravna zamjena metalnim cjevovodima. KEY WORDS plastic pipelines corrosion resistance operational costs shipbuilding marine industry KLJUČNE RIJEČI plastični cjevovodi otpornost na koroziju operativni troškovi brodogradnja pomorska industrija

The use of plastic pipelines allows to increase the efficiency of pipeline systems of a ship due to their extraordinary corrosion resistance, a decrease of maintenance frequency, a reduction of total operational costs, and opens a new direction in the use of plastics. The weight of plastic pipelines is considerably less than that of metal ones, which allows to decrease the weight of the vessel by several tons and to increase, due to this, its maneuverability, as well as to decrease fuel consumption, ensure easier passing of all the necessary weight regulations, and optimize vessel operation. The process of installation work performance provides for a possibility of a pre-assembly of structures, the execution of welded, glued, mechanical joints with the help of a minimum set of tools in a confined space, and decreases by 50% the time of pipeline laying on the ship. Moreover, plastic pipe manufacturing creates five times less carbon dioxide emissions, which makes it a significant envi...

MEGARA (Multi-Espectrógrafo en GTC de Alta Resolución para Astronomía) is the future intermediate-resolution optical Integral-Field Unit (IFU) and Multi-Object Spectrograph (MOS) of the 10.4m GTC telescope. The instrument can be used to observe either a contiguous (100% filling factor) field-of-view of 12.5×11.3 arcsec^{2} or 92 objects anywhere in a 3.5×3.5 arcmin^{2} field patrolled by robotic actuactors attached to optical-fiber minibundles, respectively in its IFU and MOS modes. The MEGARA Consortium is led by the Universidad Complutense de Madrid (UCM, Spain) and also includes the Instituto Nacional de Astrofísica, óptica y Electrónica (INAOE, Mexico), the Instituto de Astrofísica de Andalucía (IAA-CSIC, Spain) and the Universidad Politécnica de Madrid (UPM, Spain). The instrument passed its Critical Design Review (CDR) on late 2014 and is currently in construction phase with a planned date for the start of operations at GTC on early 2017. In this paper we summarize the main ch...

A model of user behaviour based automated planning is introduced in this work. The behaviour of users of web interactive systems can be described in term of a planning domain encapsulating the timed actions patterns representing the intended user profile. The user behaviour recognition is then posed as a planning problem where the goal is to parse a given sequence of user logs of the observed activities while reaching a final state. A general technique for transforming a timed finite state automata description of the behaviour into a numerical parameter planning model is introduced. Experimental results show that the performance of a planning based behaviour model is effective and scalable for real world applications. A major advantage of the planning based approach is to represent in a single automated reasoning framework problems of plan recognitions, plan synthesis and plan optimisation.

In this paper RDPPLan, a model for planning with quantitative resources specified as numerical intervals, is presented. Nearly all existing models of planning with resources require to specify exact values for updating resources modified by actions execution. In other words these models cannot deal with more realistic situations in which the resources quantities are not completely known but are bounded by intervals. The RDPPlan model allow to manage domains more tailored to real world, where preconditions and effects over quantitative resources can be specified by intervals of values, in addition mixed logical/quantitative and pure numerical goals can be posed. RDPPlan is based on non directional search over a planning graph, like DPPlan, from which it derives, it uses propagation rules which have been appropriately extended to the management of resource intervals. The propagation rules extended with resources must verify invariant properties over the planning graph which have been proven by the authors and guarantee the correctness of the approach. An implementation of the RDPPlan model is described with search strategies specifically developed for interval resources.

Scheduling under uncertainty is an area of interest in artificial intelligence. We study the problem of Dynamic Controllability (DC) of Disjunctive Temporal Networks with Uncertainty (DTNU), which seeks a reactive scheduling strategy to satisfy temporal constraints in response to uncontrollable action durations. We introduce new semantics for reactive scheduling: Time-based Dynamic Controllability (TDC) and a restricted subset of TDC, R-TDC. We present a tree search approach to determine whether or not a DTNU is R-TDC. Moreover, we leverage the learning capability of a Graph Neural Network (GNN) as a heuristic for tree search guidance. Finally, we conduct experiments on a known benchmark on which we show R-TDC to retain significant completeness with regard to DC, while being faster to prove. This results in the tree search processing fifty percent more DTNU problems in R-TDC than the state-of-the-art DC solver does in DC with the same time budget. We also observe that GNN tree search guidance leads to substantial performance gains on benchmarks of more complex DTNUs, with up to eleven times more problems solved than the baseline tree search.

Scheduling in the presence of uncertainty is an area of interest in artificial intelligence due to the large number of applications. We study the problem of dynamic controllability (DC) of disjunctive temporal networks with uncertainty (DTNU), which seeks a strategy to satisfy all constraints in response to uncontrollable action durations. We introduce a more restricted, stronger form of controllability than DC for DTNUs, time-based dynamic controllability (TDC), and present a tree search approach to determine whether or not a DTNU is TDC. Moreover, we leverage the learning capability of a message passing neural network (MPNN) as a heuristic for tree search guidance. Finally, we conduct experiments for which the tree search shows superior results to state-of-the-art timed-game automata (TGA) based approaches. We observe that using an MPNN for tree search guidance leads to a significant increase in solving performance and scalability to harder DTNU problems.

Planning for Autonomous Unmanned Ground Vehicles (AUGV) is still a challenge, especially in difficult, off-road critical situations. Automatic planning can be used to reach mission objectives, to perform navigation or maneuvers. Most of the time, the problem consists in finding a path from a source to a destination, while satisfying some operational constraints. In a graph without negative cycles, the computation of the single-pair shortest path from a start node to an end node is solved in polynomial time. Additional constraints on the solution path can however make the problem harder to solve. That becomes the case if we require the path to pass by a few mandatory nodes without requiring a specific order of visit. The complexity grows exponentially as we require more nodes to visit. In this paper, we focus on shortest path search with mandatory nodes on a given connected graph. We propose a hybrid model that combines a constraintbased solver and a graph convolutional neural network to improve search performance. Promising results are obtained on realistic scenarios.

The current INFRAWEBS European research project aims at developing ICT framework enabling software and service providers to generate and establish open and extensible development platforms for Web Service applications. One of the concrete project objectives is developing a full-life-cycle software toolset for creating and maintaining Semantic Web Services (SWSs) supporting specific applications based on Web Service Modelling Ontology (WSMO) framework. According to WSMO, functional and behavioural descriptions of a SWS may be represented by means of complex logical expressions (axioms). The paper describes a specialized user-friendly tool for constructing and editing such axioms - INFRAWEBS Axiom Editor. After discussing the main design principles of the Editor, its functional architecture is briefly presented. The tool is implemented in Eclipse Graphical Environment Framework and Eclipse Rich Client Platform.

Abstract. 1 The conventional wisdom and practice is that planning and scheduling tasks are solved separately using different methods and approaches. However, recent development in industrial planning and scheduling demands for mixing both tasks to allow modelling of wider class of problems. The purpose of this paper is to present a framework for mixing planning and scheduling tasks within single system. We analyse traditional views of planning and scheduling and we highlight the drawbacks of separating both tasks when applied to modelling complex process environments. We give some real-life examples where the mixed approach helps to model the problems and we propose a generic framework for such mixture. We also argue for using constraint programming as underlying solving technology and, finally, we describe some constraint models based on the proposed framework. Although, we concentrate on planning and scheduling in complex process environments we believe that the results contribute...

1 The conventional wisdom and practice is that planning and scheduling tasks are solved separately using different methods and approaches. However, recent development in industrial planning and scheduling demands for mixing both tasks to allow modelling of wider class of problems. The purpose of this paper is to present a framework for mixing planning and scheduling tasks within single system. We analyse traditional views of planning and scheduling and we highlight the drawbacks of separating both tasks when applied to modelling complex process environments. We give some real-life examples where the mixed approach helps to model the problems and we propose a generic framework for such mixture. We also argue for using constraint programming as underlying solving technology and, finally, we describe some constraint models based on the proposed framework. Although, we concentrate on planning and scheduling in complex process environments we believe that the results contribute to both planning and scheduling communities in general.

Parallel planning is a paradigm that provides interesting efficiency improvements in the field of classical AI planning and it is one of the key components of successful SAT-based planers. Popularized by the Graphplan algorithm, it provides more structure information about the plan for a plan executor when compared to the traditional sequential plan, which is one of the crucial facts of its usability in real-life scenarios and applications. Our latest research shows that different semantics can be used for parallel planning, while deciding which semantic to use for given application can have significant influence on both practical planning domain modeling and also on the computational efficiency of searching for a plan.

We present a software tool that is able to automatically translate an NP problem into a STRIPS problem such that the former problem has a solution iff the latter has one, a solution for the latter can be transformed into a solution for the former, and all this can be done efficiently. Moreover, the tool is built such that it only produces problems that belong to a fragment of STRIPS that is solvable in non-deterministic polynomial time, a fact that guarantees that the whole approach is not an overkill. This tool has interesting applications. For example, with the advancement of planning technology, it can be used as an off-the-shelf method to solve general NP problems with the help of planners and to automatically generate benchmark problems of known complexity in a systematic and controlled manner. Another interesting contribution is related to the area of Knowledge Engineering in which one of the goals is to devise automatic methods for using the available planning technology to s...

The use of Knowledge Engineering (KE) processes to analyze and configure domains in automated planning is becoming more appealing since it was noticed that this issue could make a difference to solve real problems. The contrast between a generic domain independent approach, taken as canonical in AI, and alternative processes that include knowledge engineering-eventually adding specific knowledge-has been discussed by Computer and Engineering communities. A big impact has been noticed mainly in the early phase of requirement analysis when KE approach is normally introduced. Requirement analysis is responsible for carrying out the Knowledge modeling of both problem and work domains, which is a key issue to guide different planner algorithms to come out with efficient solutions. Also, there is the scalability issue that appear in most real problems. To face that, hierarchical methods played an important hole in the history of planning and inspired several solutions since the proposal of NONLIN in the 70's. Since then, the idea of associating hierarchical relational nets with partial ordered actions has prevailed when large systems were considered. However, there is still a gap between the hierarchical approach and the state of art of requirements analysis to allow features anticipated by KE approach to really appear in the requirements of a planning process. This paper proposes a pathway to solve this gap starting with requirements elicitation represented first in the conventional semi-formal (diagrammatic) language-UML-that is translated to Hierarchical Petri Nets (HPNs) by a new enhanced algorithm. The proposed process was installed in a software tool-developed by one of the authors-that analyzes the performance of the KE planning model: itSIMPLE (Integrated Tools Software Interface for Modeling Planning Environment). This tool was initially designed to use classic Place/Transition nets and an old version of UML (2.1). It is now enhanced to use UML 2.4 and a hierarchical Petri Net extension, also developed by the authors. Realistic examples illustrate the process which is now being applied to larger problems related to the manufacturing of car sequencing domain, one of challenge of ROADEF 2005 (French Operations Research & Decision Support Society). Finally, we consider the possibility to introduce another approach to the KE process by using KAOS (Keep All Object Satisfied) to make the planning design more accurate. ✩ No author associated with this paper has disclosed any potential or pertinent conflicts which may be perceived to have impending conflict with this work.

A great effort has been made nowadays in the area of Artificial Intelligence for defining reliable automated planning systems that can be flexible and integrated. That leads to the need of a systematic design process, in which the initial phases are not neglected. In this work we show the importance of these initial design phases applied to planning systems taking advantage of a parallel characterization of domain and problem. Thus, we propose a modeling and analyzing environment, called itSIMPLE 2.0 , which specially contemplates viewpoint concepts from the Requirements Engineering, as well as general concepts from Knowledge Engineering and the Engineering Design. In this environment, it's possible to integrate different system representations and analysis like UML, XML, Petri Nets, and PDDL, in order to obtain a well-formed model which represents the viewpoints of important classes of participants. This integration, which is totally XML based, aims to provide the potential of each language and analysis engine during the design process of an automated planning system for real applications. A case study is included to the management of transportation of oil in the port of São Sebastião in São Paulo province. Thus, this well-known and complex example is used to compare the facility of announcing and structuring the problem in the early phase of the design process.

In this paper, we investigate how knowledge acquired during a plan analysis phase that follows model design affects planning performance. We describe a postdesign framework that combines a knowledge engineering tool and a virtual prototyping environment for the analysis and simulation of plans. Our framework demonstrates that post-design analysis supports the discovery of missing requirements and guides the model refinement cycle. We present two case studies using benchmark domains and eight state-of-the-art planners. Our results demonstrate that significant improvements in plan quality and an increase in planning speed of up to three orders of magnitude can be achieved through a careful post-design process. We argue that such a process is critical for deployment of planning technology in real-world applications.