MK

Resources

Below are some tangible outputs of my research, including software, conceptual models, domain models, and datasets, and lecture notes (the webpage is a work in progress).

Software

  • SH Planning System
    • Purpose: SH is an AI planning system based on state-based Hierarchical Task Network (HTN) planning. As a domain-independent planner, it can handle various planning problem descriptions without requiring modifications to its underlying algorithms.
    • Usage: SH can be used by researchers, practitioners, students, and anyone who needs a modern, robust HTN planning system. Its versatility makes it suitable for a wide range of planning problems across multiple domains.
    • [Webpage][Code] [Documentation] [Paper]
  • PlanX Toolbox
    • Purpose: PlanX Toolbox is a collection of planning services for building and integrating AI planning systems. It includes one messaging component and multiple planning services, such as modelling, visualisation, system monitoring, parsing, converting, solving, and plan validation.
    • Usage: The toolbox is aimed at researchers and developers who need a simplified way to develop advanced AI planning systems. It is well-suited for composing AI planning systems that need to be integrated into larger software architectures.
    • [Code] [Documentation] [Demo] [Paper]
  • HDDL2HPDL Converter
    • Purpose: HDDL2HPDL Converter is a tool that converts HTN domain models and problem instances specified in the Hierarchical Domain Definition Language (HDDL) into the Hierarchical Planning Definition Language (HPDL), the specification language supported by the SH Planning System.
    • Usage: The tool is intended for researchers and developers who want to use the SH Planning System for HTN planning problems specified in HDDL.
    • [Code][Paper]

Platforms

  • PSpace
    • Purpose: PSpace is a centralised platform that aggregates AI planning software, such as planners and domain modellers. It allows users to search for planning tools and share information on developed software. The platform aims at fostering collaboration among AI planning researchers and practitioners.
    • Usage: PSpace is designed for researchers, practitioners, and students seeking planning software tailored to their specific tasks, or for sharing information about available tools.
    • [Website] [Documentation] [Paper]
  • DSpace
    • Purpose: DSpace is a centralised platform for storing, validating, and viewing planning domain models. It provides a framework for documenting domain models through value cards, enabling transparency and standardisation across the planning community.
    • Usage: DSpace supports researchers, practitioners, and students in searching for or sharing planning domain models. The platform stores and validates domain models specified in PDDL, HPDL, and HDDL.
    • [Code] [Documentation] [Paper]

Conceptual Models

  • PlanXFlow
    • Purpose: PlanXFlow is a software development life cycle designed for engineering AI planning systems. PlanXFlow provides a structured, waterfall-like yet iterative approach tailored to the unique challenges in AI planning.
    • Features: This life cycle consists of ten phases, guiding practitioners and researchers from the conceptualisation of AI planning systems to their deployment and beyond. Its core features include adaptability, scaling, transparency, and future-proofing.
    • [Webpage] [Paper]
  • D2F
    • Purpose: D2F is a conceptual framework characteristing features of planning domain models. It can impact the design, development, and applicability of AI planning systems in real-world settings.
    • Features: The framework provides a common terminology, feature categories, and a broad range of planning features suitable for real-world planning domains.
    • [Model] [Paper]
  • Planomics
    • Purpose: Planomics is a structured body of knowledge for identifying, classifying, and reasoning about functionalities of AI planning systems. These functionalities are basic, distinct blocks that, when interconnected, form an ecosystem, that is, an advanced AI planning system.
    • Features: Planomics includes a wide spectrum of functionalities, including modelling and parsing planning problems, plan generation and execution, data management, system management, and system monitoring. Furthermore, Planomics categorises planning functionalities into classes, indicating their role within the ecosystem. It also positions these functionality classes in terms of their contribution to the functional richness and evolution of the AI planning system capability.
    • [Model] [Paper]

Lecture Notes

  • Introduction to AI Planning
    • Description: These notes cover lectures at the University of Stuttgart and provide an introduction to key concepts and techniques in AI Planning, also known as Automated Planning. The notes begin by introducing the state model and then explore classical planning, the foundational form of planning, and present fundamental algorithms for solving such problems. Subsequently, we examine planning as a constraint satisfaction problem, outline the mapping process, and describe an approach to solving such problems. The most extensive section is dedicated to Hierarchical Task Network (HTN) planning, one of the most widely used and powerful planning techniques in the field. The lecture notes end with a bonus chapter on the Planning Domain Definition (PDDL) Language, the de facto standard syntax for representing non-hierarchical planning problems.
    • Authors: Marco Aiello and Ilche Georgievski
    • [Notes]

Domain Models

TBD