Paper Type

ERF

Abstract

Digitalization drives manufacturers to develop flexible and intelligent production processes, yet manual pre-assembly remains difficult to automate due to high product variability. This creates demand for intelligent worker assistance systems. While augmented reality (AR) and artificial intelligence (AI) have shown potential for improving manual assembly, integrated AR-AI architectures connected to existing IT infrastructure remain underexplored. This study derives five design requirements from a systematic literature review and develops a four-component architecture using a design science approach: a pick-by-vision system with spatial calibration, a YOLO11n-based object detection and confirmation-based step validation pipeline, an MQTT-based data exchange layer, and a virtual material guidance module. We instantiate the architecture in a drone pre-assembly smart factory at a university-affiliated research facility using a Meta Quest 3 headset. Using the Framework for Evaluation in Design Science (FEDS), we demonstrate technical feasibility and derive provisional design principles for human-centered assembly assistance.

Paper Number

1431

Comments

NEXTTRANS

Download

Share

COinS
 
Aug 15th, 12:00 AM

AI-/AR-Architecture for Operation Processes in Smart Factories

Digitalization drives manufacturers to develop flexible and intelligent production processes, yet manual pre-assembly remains difficult to automate due to high product variability. This creates demand for intelligent worker assistance systems. While augmented reality (AR) and artificial intelligence (AI) have shown potential for improving manual assembly, integrated AR-AI architectures connected to existing IT infrastructure remain underexplored. This study derives five design requirements from a systematic literature review and develops a four-component architecture using a design science approach: a pick-by-vision system with spatial calibration, a YOLO11n-based object detection and confirmation-based step validation pipeline, an MQTT-based data exchange layer, and a virtual material guidance module. We instantiate the architecture in a drone pre-assembly smart factory at a university-affiliated research facility using a Meta Quest 3 headset. Using the Framework for Evaluation in Design Science (FEDS), we demonstrate technical feasibility and derive provisional design principles for human-centered assembly assistance.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.