What question did this study set out to answer?

The paper aims to address the gap in manufacturing digital transformation between advanced technology and actual worker experience.

March 1, 2026Open Access

The Last Mile of Manufacturing DX

Key Points

The paper aims to address the gap in manufacturing digital transformation between advanced technology and actual worker experience.
Analysis of existing smart factory infrastructures and digital transformation challenges.
Development of an on-device XR system for real-time AI inference.
Implementation of dual NPU architecture for high AI processing capability.
Use of triple sensor fusion for spatial accuracy in factory settings.
Deployment of a hands-free multimodal interface for user interaction.
Reduction of training time by 50% in shipyard applications.
70% decrease in material waste observed from technology use.
Demonstration of zero perceptual delay for motion responses in the XR interface.
Successful operational implementation at 5 major shipyards.

Abstract

Technical Whitepaper | For Manufacturing Engineers The Last Mile of Manufacturing DX:You Invested in Smart Factory Infrastructure. Why Hasn't the Shop Floor Changed? MES is installed. IoT sensors are deployed. Dashboards are running. Yet the worker at the machine still carries a paper drawing. This is the "last mile problem" of manufacturing DX. Equipment and systems have gone digital, but the most critical interface — the worker's eyes and hands — remains analog. This whitepaper examines how to close this last mile: an on-device XR system engineering architecture that delivers AI inference results directly into the worker's field of view, in real time, without network dependency. Key topics covered:- Diagnosing the Last Mile Gap: Why DX investments fail to reach workers- Dual NPU Architecture: 125 TOPS AI processing in a belt-worn device- Triple Sensor Fusion: 2.47cm spatial accuracy in harsh factory environments- Hands-Free Multimodal Interface: Gesture, voice, and gaze control with gloves and tools- Sub-21ms Motion-to-Photon Latency: Zero perceptual delay for 8-hour shifts- 6-Stage AI Model Optimization: Server-room AI compressed to 120MB on-device- Field Evidence: Deployed at 5 HD Hyundai Group shipyards — 50% training time reduction, 70% material waste reduction- Zero-Cloud Security Architecture: All data processed and discarded on-device This is not theory. The underlying technology is operational across 5 of the world's largest shipyards under HD Hyundai Group. Published by MetaVu Co., Ltd. (www.metavu.io)Contact: support@metavu.io

Read Full Paperexternally

Bookmark

View Full Paper

Cite This Study

ROH et al. (Wed,) studied this question.

synapsesocial.com/papers/69a3d7eeec16d51705d2e65c https://doi.org/https://doi.org/10.5281/zenodo.18795154

Bookmark

View Full Paper