工業4.0條件下的機器設備因接收數位訊號直接控制，故可做到完美的動作。但於人機協作的條件發生時，人員雖可以聽命行事，人卻因擁有主動意識，使其動作存在許多不確定性，潛在可能造成對協同任務的不預期影響。因此，執行人機協同時，智慧製造系統須能即時認知人員動作，並根據該項協同任務的工作目標來同步協調人機間之動作直到任務完成。於是，本研究將探討以雙深度攝影機對人員做即時肢體動作參數的擷取、融合與判讀，藉此改善單台深度攝影機於動作範圍的侷限，並取得精度較高的肢體動作參數，以供規劃人機協同任務時之參考。

Under the definition of Industry 4.0, all production facilities are communicated by direct digital signals, and so connectivities among collaborative works can be perfectly achieved. However, in the scenario, human workers were ruled out. In fact, man-machine collaboration is inevitable even in a manless factory. Human are still the effective resource for certain delicate tasks, despite the nature of their uncertainties were considered the trade-off of the employment. A logically perfect collaboration should take human motions into consideration and plan along with the machine’s operations through out the tasks. Therefore, in this research, a dual depth camera scheme is adopted to provide a wider range of motion capture and inspection. The dual data will then fusion into a stick figure human with higher precision to support the planning of an intelligent man-machine collaboration task.