Current Issue

Volume 10 - Issue 06 (November - December 2026)

 

Title: Gaze-Guided Wearable AR, Multimodal AI, and Digital Twins for Smart Factory Inspection
Authors: Yu-Ching Lin
Source: International Journal of Latest Research in Engineering and Management, pp 01 - 06, Vol 10 - No. 06, 2026
Abstract: Routine equipment inspection on factory floors demands hands-free human-machine interfaces that allow technicians to access machinery diagnostics and technical documentation safely while keeping both hands free. This paper presents a practical four-tier system architecture that integrates wearable optical see-through augmented reality (AR) smart glasses, multimodal artificial intelligence (AI), an Advanced Message Queuing Protocol (AMQP) broker, and a web-based 3D digital twin for smart factory equipment monitoring. On the wearable client, binocular eye-gaze tracking is combined with an inverse center-crop coordinate mapping and a 3.0 s fixed-anchor dwell trigger to crop 512 × 512 regions of interest (ROIs) accurately from 1600 × 1200 camera frames. On the server tier, a dual-mode vision-language model (VLM) distinguishes industrial machinery from technical text and links recognized equipment identifiers to a canonical asset catalog to provide reliable reference telemetry alongside document translation. In parallel, a streaming 16 kHz voice pipeline combines neural voice activity detection (VAD), domain-prompted speech recognition, a 600 ms half-duplex echo guard, and a six-tool whitelist agent. Finally, a WebGL/Three.js 3D digital twin visualizes real-time equipment states, peer-to-peer inspection sharing, and cumulative gaze attention. Prototype verification on a physical pump skid testbed confirms reliable end-to-end operation.
Download Full Article