Artificial Intelligence Planning Research Articles

Smart Enough to Talk With Us? Foundations and Challenges for Dialogue Capable AI Systems

Smart Enough to Talk With Us? Foundations and Challenges for Dialogue Capable AI Systems

Cooperative Multi-Agent Planning

Cooperative multi-agent planning (MAP) is a relatively recent research field that combines technologies, algorithms, and techniques developed by the Artificial Intelligence Planning and Multi-Agent Systems communities. While planning has been generally treated as a single-agent task, MAP generalizes this concept by considering multiple intelligent agents that work cooperatively to develop a course of action that satisfies the goals of the group. This article reviews the most relevant approaches to MAP, putting the focus on the solvers that took part in the 2015 Competition of Distributed and Multi-Agent Planning, and classifies them according to their key features and relative performance.

Autonomous Collision-Free Navigation of Microvehicles in Complex and Dynamically Changing Environments.

Self-propelled micro- and nanoscale robots represent a rapidly emerging and fascinating robotics research area. However, designing autonomous and adaptive control systems for operating micro/nanorobotics in complex and dynamically changing environments, which is a highly demanding feature, is still an unmet challenge. Here we describe a smart microvehicle for precise autonomous navigation in complicated environments and traffic scenarios. The fully autonomous navigation system of the smart microvehicle is composed of a microscope-coupled CCD camera, an artificial intelligence planner, and a magnetic field generator. The microscope-coupled CCD camera provides real-time localization of the chemically powered Janus microsphere vehicle and environmental detection for path planning to generate optimal collision-free routes, while the moving direction of the microrobot toward a reference position is determined by the external electromagnetic torque. Real-time object detection offers adaptive path planning in response to dynamically changing environments. We demonstrate that the autonomous navigation system can guide the vehicle movement in complex patterns, in the presence of dynamically changing obstacles, and in complex biological environments. Such a navigation system for micro/nanoscale vehicles, relying on vision-based close-loop control and path planning, is highly promising for their autonomous operation in complex dynamic settings and unpredictable scenarios expected in a variety of realistic nanoscale scenarios.

A compiler for MSVL and its applications

In this paper, we present a method for implementing a compiler called MC for Modeling, Simulation and Verification Language (MSVL) based on LLVM. MC accepts a well formed MSVL program as input and generates an executable binary code. Different from other compilers, MC can be used not only to compile programs but also to model and verify programs. The details of implementation including the architecture design, lexical, syntactic and semantic analysis, as well as preprocessing and scheduling algorithms. Besides, we show MC can be utilized as an infrastructure for Artificial Intelligence (AI) planning. Several examples are given to show applications of MC in modeling and verifying programs, as well as AI planning.

Design and evaluation of planning and mathematical models for generating learning paths

AbstractEducation research has emphasized the need to develop instructional design tools to facilitate the generation of learning paths for students. Learning paths are important because they enable the personalization and optimization of the learning process. In this work, we present a flexible conceptual framework that allows the representation of curricula information as Artificial Intelligence Planning and Mathematical Programming models to facilitate the generation of learning paths by domain independent algorithms. The resulting models consider a rich set of properties from the education domain, like hierarchical learning structures, enabling conditions, temporal actions, mandatory activities, quality accumulation functions, and metric information. We show that the proposed mathematical models return optimal solutions very efficiently if we relax the total ordering constraints of learning paths. These relaxations allow evaluating greedy planning algorithms to identify the properties from the models that increase the complexity of solution synthesis. We expect that the results of this research can be helpful to education researchers and computer scientists in the quest of scalable systems that capture more flexible standards to model learning and compute more informed learning paths for students.

Stan4: A Hybrid Planning Strategy Based on Subproblem Abstraction

Planning domains often feature subproblems such as route planning and resource handling. Using static domain analysis techniques, we have been able to identify certain commonly occurring subproblems within planning domains, making it possible to abstract these subproblems from the overall goals of the planner and deploy specialized technology to handle them in a way integrated with the broader planning activities. Using two such subsolvers our hybrid planner, stan4, participated successfully in the Fifth International Conference on Artificial Intelligence Planning and Scheduling (AIPS'00) planning competition.

Introduction to the special issue on constraint satisfaction for planning and scheduling

AbstractThe areas of Artificial Intelligence planning and scheduling have seen important advances thanks to the application of constraint satisfaction models and techniques. Especially, solutions to many real-world problems need to integrate plan synthesis capabilities with resource allocation, which can be efficiently managed by using constraint satisfaction techniques. Constraint satisfaction plays an important role in solving such real life problems, and integrated techniques that manage planning and scheduling with constraint satisfaction are particularly useful.

Symbolic On-the-Fly Synthesis in Supervisory Control Theory

This paper presents an efficient synthesis algorithm and its proof of correctness for computing the controllable, nonblocking, and minimally restrictive supervisor in the supervisory control theory. Conventional synthesis algorithms are based on backward reachability computations, where blocking and uncontrollable states are iteratively found by searching the entire state space several times until a fixed point is reached. Many unnecessary states may be visited in this kind of searching. In this paper, we present an alternative synthesis algorithm based on forward reachability, where a number of synthesis steps are performed during the reachability computations. This approach is inspired from the search techniques in Artificial Intelligence (AI) planning. To handle large-scale problems, the algorithm performs the computations symbolically based on binary decision diagrams . The algorithm has been developed, implemented, and applied to several large-scale benchmarks. It is shown that, on average, the on-the-fly algorithm is more efficient than the conventional synthesis algorithms, in particular for problems with many uncontrollable states.

Apprenticeship Scheduling for Human-Robot Teams

Resource optimization and scheduling is a costly, challenging problem that affects almost every aspect of our lives. One example that affects each of us is health care: Poor systems design and scheduling of resources can lead to higher rates of patient noncompliance and burnout of health care providers, as highlighted by the Institute of Medicine (Brandenburg et al. 2015). In aerospace manufacturing, every minute re-scheduling in response to dynamic disruptions in the build process of a Boeing 747 can cost up to $100.000. The military is also highly invested in the effective use of resources. In missile defense, for example, operators must =solve a challenging weapon-to-target problem, balancing the cost of expendable, defensive weapons while hedging against uncertainty in adversaries’ tactics. Researchers in artificial intelligence (AI) planning and scheduling strive to develop algorithms to improve resource allocation. However, there are two primary challenges. First, optimal task allocation and sequencing with upper and lower-bound temporal constraints (i.e., deadlines and wait constraints) is NP-Hard (Bertsimas and Weismantel 2005). Approximation techniques for scheduling exist and typically rely on the algorithm designer crafting heuristics based on domain expertise to decompose or structure the scheduling problem and prioritize the manner in which resources are allocated and tasks are sequenced (Tang and Parker 2005; Jones, Dias, and Stentz 2011). The second problem is this aforementioned reliance on crafting clever heuristics based on domain knowledge. Manually capturing domain knowledge within a scheduling algorithm remains a challenging process and leaves much to be desired (Ryan et al. 2013). The aim of my thesis is to develop an autonomous system that 1) learns the heuristics and implicit rules-of-thumb developed by domain experts from years of experience, 2) embeds and leverages this knowledge within a scalable resource optimization framework, and 3) provides decision support in a way that engages users and benefits them in their decision-making process. By intelligently leveraging the ability of humans to learn heuristics and the speed of modern computation, we can improve the ability to coordinate resources in these time and safety-critical domains.

A STUDY ON ARTIFICIAL INTELLIGENCE PLANNING

Human planning is an act of approaching goals. Its nature is explicit that is humans may proceed with prior planning and without planning. We mostly plan out things when the situation is too demanding. Artificial intelligence planning is the intelligence of planning exhibited by machines or software. It is an academic field of study which studies goal of creating intelligence to plan virtually for obtaining optimum results by problem formulation. The initial state, the set of possible actions, goal and path cost functions are main attributes of AI planning. This paper is a concise study of planning algorithms in Artificial Intelligence.

Symbolic discrete-time planning with continuous numeric action parameters for agent-controlled processes

In industrial domains such as manufacturing control, a trend away from centralized planning and scheduling towards more flexible distributed agent-based approaches could be observed over recent years. To be of practical relevance, the local control mechanisms of the autonomous agents must be able to dependably adhere and dynamically adjust to complex numeric goal systems like business key performance indicators in an economically beneficial way. However, planning with numeric state variables and objectives still poses a challenging task within the field of artificial intelligence (AI).In this article, a new general-purpose AI planning approach is presented that operates in two stages and extends existing domain-independent modeling formalisms like PDDL with continuous (i.e., infinite-domain) numeric action parameters, which are currently still unsupported by state-of-the-art AI planners. In doing so, it enables the solution of mathematical optimization problems at the action level of the planning tasks, which are inherent to many real-world control problems. To deal with certain difficulties concerning reliable and fast detection of action applicability that arise when planning with real-valued action parameters, the implemented planner allows resorting to an adjustable “satisficing” strategy by means of partial execution and subsequent repair of infeasible plans over the course of time. The functioning of the system is evaluated in a multi-agent simulation of a shop floor control scenario with focus on the effects the possible problem cases and different degrees of satisficing have on attained plan quality and total planning time. As the results demonstrate the basic practicability of the approach for the given setting, this contribution constitutes an important step towards the effective and dependable integration of complex numeric goal systems and non-linear multi-criteria optimization tasks into autonomous agent-controlled industrial processes in a reusable, domain-independent way.

Narrative Generation in Entertainment: Using Artificial Intelligence Planning

From the field of artificial intelligence (AI) there is a growing stream of technology capable of being embedded in software that will reshape the way we interact with our environment in our everyday lives. This ‘AI software’ is often used to tackle more mundane tasks that are otherwise dangerous or meticulous for a human to accomplish. One particular area, explored in this paper, is for AI software to assist in supporting the enjoyable aspects of the lives of humans. Entertainment is one of these aspects, and often includes storytelling in some form no matter what the type of media, including television, films, video games, etc. This paper aims to explore the ability of AI software to automate the story-creation and story-telling process. This is part of the field of Automatic Narrative Generator (ANG), which aims to produce intuitive interfaces to support people (without any previous programming experience) to use tools to generate stories, based on their ideas of the kind of characters, intentions, events and spaces they want to be in the story. The paper includes details of such AI software created by the author that can be downloaded and used by the reader for this purpose. Applications of this kind of technology include the automatic generation of story lines for ‘soap operas’.

Artificial Intelligence Planning for AUV Mission Control

Autonomous intelligent robotic systems are becoming increasingly important in a wide range of applications. Many of these application contexts are in physical situations out of human reach, so that a robot, or team of robots, must be capable of operating for long periods without human intervention. This requires a strategic planning capability as well as an ability to interpret and adapt to unexpected events. In this paper we describe progress towards developing such a capability in the context of underwater oilfield operations.

Interpreting Natural Language Instructions Using Language, Vision, and Behavior

We define the problem of automatic instruction interpretation as follows. Given a natural language instruction, can we automatically predict what an instruction follower, such as a robot, should do in the environment to follow that instruction? Previous approaches to automatic instruction interpretation have required either extensive domain-dependent rule writing or extensive manually annotated corpora. This article presents a novel approach that leverages a large amount of unannotated, easy-to-collect data from humans interacting in a game-like environment. Our approach uses an automatic annotation phase based on artificial intelligence planning, for which two different annotation strategies are compared: one based on behavioral information and the other based on visibility information. The resulting annotations are used as training data for different automatic classifiers. This algorithm is based on the intuition that the problem of interpreting a situated instruction can be cast as a classification problem of choosing among the actions that are possible in the situation. Classification is done by combining language, vision, and behavior information. Our empirical analysis shows that machine learning classifiers achieve 77% accuracy on this task on available English corpora and 74% on similar German corpora. Finally, the inclusion of human feedback in the interpretation process is shown to boost performance to 92% for the English corpus and 90% for the German corpus.

Automatic Composition of Heterogeneous Models Based on Semantic Web Services

As an important function of a distributed decision support system, model composition aims to aggregate model functions to solve complex decision problems. Most existing methods on model composition only apply to the models which have the same type of input and output data so that they can be linked together directly. Those methods are inadequate for the heterogeneous models, since a heterogeneous model may have different types of input and output data that are represented in either qualitative or quantitative manner. This paper aims to address the problem of heterogeneous model composition by employing the techniques based on semantic web services and artificial intelligence planning. In this paper, the heterogeneous model composition problem is converted to the problem of planning in nondeterministic domains under partial observability. An automatic composition method is presented to generate the composite model based on the planning as model checking technique. The experiment results are also presented in this paper to show the feasibility and capability of our approach in dealing with the complex problems involving heterogeneous models.

Applying CBR Over an AI Planner for Dynamic Web Service Composition

Web services have emerged as a major technology for deploying automated interactions between distributed and heterogeneous applications. The main advantage of Web services composition is the possibility of creating value-added services by combining existing ones to achieve customized tasks. How to combine these services efficiently into an arrangement that is both functionally sound and architecturally realizable is a very challenging topic that has founded a significant research area within computer science. A great deal of recent Web-related research has concentrated on dynamic Web service composition. Most of proposed models for dynamic composition use semantic descriptions of Web services through the construction of domain ontology. In this paper, we present our approach to dynamically produce composite services. It is based on the use of two Artificial Intelligence (AI) techniques: Case-Based Reasoning (CBR) and AI planning. Our motivating scenario concerns a national system for the monitoring of childhood immunization.

A knowledge-based architecture for the management of patient-focused care pathways

There is need for software systems in order to coordinate the activities of healthcare professionals involved in the treatment of a patient, aligning the care delivery around already existing Clinical Practice Guidelines (CPGs). This is being carried out in care organizations by implementing integrated Care Pathways (CPs). Nonetheless, the generation of these care pathways is not trivial, and multiple barriers exist for their development and enactment. In this paper, a knowledge-based architecture is presented that, by means of Knowledge Engineering methods and Artificial Intelligence Planning and Scheduling (AI P&S) techniques, is able to automatically generate these care pathways from a computer-interpretable representation of CPGs, tackling some of these barriers. Firstly, these techniques consider the patient profile, the care organization details as well as the temporal and resource constraints, implicit in a care process, in order to generate a patient-focused care pathway. Moreover, they also allow the enactment of personalized care plans in a web-based format, powered by a workflow runtime engine, thus providing an ubiquitous and interactive execution to healthcare professionals. Finally, the architecture also includes monitoring and replanning techniques in order to check the current health status of patients and adapt care plans when they do not progress as expected. For the experimental evaluation of the architecture, several tests have been carried out in order to simulate a clinical environment where different care plans were automatically executed, monitored and adapted regarding the health conditions of patients as well as the recommendations specified in a real, CPG of the paediatric oncology area. As conclusion, the proposed architecture seems to be an adequate infrastructure for supporting the automated generation as well as the interactive execution and monitoring of patient-focused care pathways.

Automatic composition of WSMO based geospatial semantic web services using artificial intelligence planning

Geospatial web services are widely utilised to disseminate spatial data and processing capability over the internet. Combining geospatial web services and generating composite web services out of atomic services can provide users with a higher level of functionality. This paper presents a solution for the automatic composition of geospatial web services. The solution relies on semantic web services technology and Artificial Intelligence (AI) planning techniques to generate composite geospatial web services. Web Service Modelling Ontology (WSMO) is selected as the underlying semantic web service framework. A neoclassical planning algorithm called Graphplan is used as the AI planner. The applicability of the proposed solution for composing web services is illustrated in a case study. The result shows that the proposed approach is well suited for the automatic composition of web services. The future requirements for adaptation of the approach in real world problems are broadly described.

Disaster planning using automated composition of semantic OGC web services: A case study in sheltering

Spatial data are crucial in disaster planning. However, because of the dynamic, urgent and uncertain nature of disasters, certain data and functionalities may be inaccessible to decision makers when they are required. Web service composition offers a possible solution whereby disaster planners can integrate spatial web services to generate new spatial data and functionalities, quickly, from existing ones. This paper proposes an automatic solution for composing OWSs (Open Geospatial Consortium Web Services) for disaster planning. A semantic annotation approach based on the Resource Description Framework (RDF) and SPARQL languages is used to describe OWSs semantically. A conceptual model for AI (Artificial Intelligence) planning is also proposed that works based on RDF and SPARQL. An AI planning algorithm was implemented based on the proposed conceptual model to compose semantic OWSs. The applicability of the proposed solution is investigated through a case study in evacuation sheltering. The case study demonstrates that the proposed automatic composition approach can enhance the efficiency of OWS integration and thereby improve the disaster management process.

Acquiring planning domain models using LOCM

AbstractThe problem of formulating knowledge bases containing action schema is a central concern in knowledge engineering for artificial intelligence (AI) planning. This paper describes Learning Object-Centred Models (LOCM), a system that carries out the automated generation of a planning domain model from example training plans. The novelty of LOCM is that it can induce action schema without being provided with any information about predicates or initial, goal or intermediate state descriptions for the example action sequences. Each plan is assumed to be a sound sequence of actions; each action in a plan is stated as a name and a list of objects that the action refers to. LOCM exploits assumptions about the kinds of domain model it has to generate, rather than handcrafted clues or planner-oriented knowledge. It assumes that actions change the state of objects, and require objects to be in a certain state before they can be executed. In this paper, we describe the implemented LOCM algorithm, the assumptions that it is based on, and an evaluation using plans generated through goal-directed solutions, through random walk, and through logging human-generated plans for the game of freecell. We analyze the performance of LOCM by its application to the induction of domain models from five domains.