基於雙物聯網與擴增實境之智慧定心量產製程人機互動決策系統

近年來，科技進步使人工智慧成為提升製程效能和生產良率的關鍵。針對光學玻璃鏡片定心製程，為降低人工經驗依賴、建立即時診斷機制及縮短調機時程，本研究開發基於雙物聯網架構與擴增實境（Augmented Reality, AR） 等技術之人機互動智慧系統，系統建立之分類模型，可精確診斷異常。接著通過雙物聯網架構串聯控制系統與人工智慧（Artificial Intelligence, AI）模型將資料即時傳輸至AR遠端控制模組，進行視覺化決策，並透過遠端控制回饋至定心機台。根據系統實測結果，在真圓度＜0.01mm、裂邊＜E0.5 的目標規格下，生產良率可從64%提升至94%，且生產時間減少29.8%，證實系統對產線製程之效益。

In recent years, advancements in technology have positioned artificial intelligence as a key enabler for improving process efficiency and production yield. In the optical glass lens centering process, conventional operations heavily rely on manual experience, lack real-time diagnostic capabilities, and require lengthy machine setup procedures. To address these limitations, this study develops a human–machine interactive intelligent system based on a dual Internet of Things (IoT) architecture integrated with Augmented Reality (AR) technologies. The proposed system establishes a classification model capable of accurately diagnosing machining anomalies. Through the dual IoT architecture, the control system and AI models are interconnected to enable real-time data transmission to a remote AR control module, where process diagnostic results and decision-making information are visualized. These decisions are further fed back to the centering machine via remote control, forming a closedloop intelligent assistance framework. Experimental validation demonstrates that, under target specifications of roundness less than 0.01 mm and edge chipping below E0.5, the production yield is improved from 64% to 94%, while total production time is reduced by 29.8%. These results confirm the effectiveness of the proposed system in enhancing process performance and production efficiency in industrial centering operations.