A Study on the Asset Administration Shell (AAS) and its Benefits to Simulations: AAS study in collaboration with Danfoss
Translated title
A Study on the Asset Administration Shell (AAS) and its Benefits to Simulations
Authors
Branco Carneiro, Tomás André ; Olori, Matthew Oghenekparobor
Term
4. term
Education
Publication year
2026
Submitted on
2026-05-29
Abstract
This thesis explores whether the Asset Administration Shell (AAS) can be used as a structured source of robot parameters for kinematic simulation, and whether this improves accuracy compared to conventional manually configured simulations. The study was carried out in collaboration with Danfoss as a laboratory case at Aalborg University, using a Universal Robots UR3 collaborative robot arm. A digital twin pipeline was developed in which the UR3 robot is described via AAS, stored on a BaSyx AAS server running in Docker, and the parameters are then used by the URSim kinematic simulator. The results are monitored and compared across the pipeline using a Streamlit web application. Using a quasi-experimental design, the performance of the real robot is compared with simulations that are either driven by AAS-based data or by manually specified parameters. The thesis presents the system architecture, implementation, experiments, results, and a discussion of limitations and future opportunities for AAS-based digital twin systems. The findings show that end-to-end parameter delivery via AAS works as intended, and that the main limitation for improving simulation fidelity is the simulator’s parameter interface, rather than the AAS architecture itself.
Denne afhandling undersøger, om Asset Administration Shell (AAS) kan bruges som en struktureret kilde til robotparametre i en kinematisk simulering, og om det kan forbedre nøjagtigheden i forhold til traditionelle, manuelt opsatte simulationer. Arbejdet er udført i samarbejde med Danfoss som et laboratoriestudie på Aalborg Universitet med en Universal Robots UR3 kollaborativ robotarm. Der er udviklet en digital tvilling-proces, hvor UR3-robotten beskrives i AAS, som lagres på en BaSyx AAS-server (kørt i Docker), og hvor data herfra anvendes i URSim-kinematik-simulatoren. Resultaterne visualiseres og sammenlignes på tværs af hele processen i en Streamlit webapplikation. Ved hjælp af et quasi-eksperimentelt design sammenlignes den fysiske robots bevægelser med simulationer, der enten er drevet af AAS-data eller manuelt konfigurerede data. Rapporten beskriver systemarkitektur, implementering, eksperimenter, resultater og en diskussion af begrænsninger samt fremtidige muligheder for AAS-baserede digitale tvillinger. Resultaterne viser, at AAS-baseret parameterlevering fungerer korrekt igennem hele systemet, og at den vigtigste begrænsning for at forbedre simulationsnøjagtigheden ligger i simulatorens måde at modtage parametre på, snarere end i AAS-arkitekturen.
[This abstract has been rewritten with the help of AI based on the project's original abstract]
Keywords