現存混合力量與姿態控制的方法,需要六個維度之力量感測器,但是六個維度之力量感測器十分的昂貴。在以前的系統中,視覺系統所扮演的角色並不是那麼的重要,它只提供控制系統所需要的參考軌跡,在本研究中,視覺系統卻是整個系統中最重要的部分。本系統運用由視覺感測器提供的資訊,遠多於經由力量感測器得到的資訊,因此,只需要較低維度的力量感測器。從所獲得的影像中搜尋使用者所發射的雷射十字光束,利用此十字光束在未知曲面上的投影來估測此曲面之法線向量,並結合力量控制對機械手臂進行定位與追蹤任務,這套系統的發展需要以下技術加以配合,視覺追蹤技術、混合任務編碼、混合力量與視覺控制理論和電腦即時控制系統。在本論文中,將上述各項技術整合後,分別設計了兩種視覺伺服架構,分別為眼在手(eye-in-hand) 和眼看手(eye-to-hand)架構,並將之實現於真實環境中。
Hybrid force and vision-based control would typically require an expensive six-axis force sensor in addition to a vision sensor. In this research, a low-cost single-axis force sensor together with a CCD camera and a laser pointer allow effective pose control of a manipulator on an unknown 3D surface. This paper presents a seemingly novel approach to autonomously control the pose of a robotic manipulator with respect to an unknown 3-D surface. Two visual servoing architectures have been proposed, one eye-to-hand and the other eye-in-hand. Our approach has been successfully validated on arbitrary 3D surfaces by performing experiments with a Mitsubishi RV-1A industrial robot. In particular, the end-effector can be driven to a set-point or a target trajectory on the surface and kept normal to the surface. A low-cost single-axis force sensor together with a CCD camera and a laser pointer are required to perform such control tasks. Compared with existing approaches, the proposed system is not only effective but also less expensive.