AAS-Planning-Metamodel

This repository defines a schema for modeling AI Planning knowledge in Asset Administration Shells (AAS). It combines four existing standards to cover the full scope of a PDDL planning problem:

• IDTA 02011 (HierarchicalStructures) for type hierarchies
• IEC 61360 (DataElementType) for predicates and state descriptions
• VDI 3682 (ProcessOperator) for capabilities/actions
• IDTA 02016 (Control Component Instances) for objects and their states

The generator tool AAS2PDDL transforms AAS models following this schema into PDDL.

Structure

• docs/AAS_STRUCTURE.md — full specification of the expected AAS structure
• templates/json/ — JSON templates for creating conforming AASX files
• examples/mini_example/ — minimal example (4 AAS, 2 actions)
• examples/mps500/ — MPS500 production system (9 AAS, 9 actions)

Example

A predicate definition in AAS:

DataElementTypes (Submodel)
└── DataElementType_On (SMC)
    ├── preferredName: "On"
    └── parameters (SMC)
        ├── param_c
        │   ├── variable: "?c"
        │   └── typeDefinitionRef → CarrierAAS/TypeDescription
        └── param_cb
            ├── variable: "?cb"
            └── typeDefinitionRef → ConveyorAAS/TypeDescription

maps to:

(:predicates (on ?c - carrier ?cb - conveyorbelt))

The PDDL predicate name is derived from the idShort (DataElementType_On → on), parameter types are resolved by following the typeDefinitionRef references to the respective TypeDescription submodels. See docs/AAS_STRUCTURE.md for the complete mapping rules.