Accelerating the Adoption of Asset Administration Shells through AI Agents

Abstract

The adoption of Industry 4.0 practices has grown worldwide, driven by the need for standardised, transparent, and interoperable data exchange. The Asset Administration Shell (AAS) provides a standardised template for asset information, enabling companies to share and integrate data across systems. However, AAS creation from existing datasheets remains a manual and time-consuming process, hindering large-scale adoption. In this paper, we propose an Artificial Intelligence (AI) agent-based approach that automates the transformation of Portable Document Format (PDF) datasheets into AAS instances, which are then serialised into AAS files. The agents extract, structure, and validate asset information against the Industrial Digital Twin Association (IDTA) guidelines to ensure compliance with industry standards. We demonstrate the approach in a use case scenario, illustrating its potential to streamline the creation of AAS files and facilitate their adoption in a manufacturing environment.

Camilo Velázquez-Rodríguez

Research Engineer

My research interests include software engineering, artificial intelligence on code and mining software repositories.