General image-based visual servoing control-law has been constrained to camera position which is fully dependent or independent to robot's velocity-screw. For camera position which is partially dependent to robot's velocity-screw, this thesis proposes a solution for control-law which uses virtual image plane.
Also this thesis proposes human-like visual servoing system for visual grasping task, which has camera position on the $1^{st}$ axis of 6 Dof. PUMA-type manipulator. Proposed system configuration provide accurate camera calibration performance with robot's 3D information. This makes it possible to predict accurate on-line image jacobian. This means that robot's trajectory can be managed reasonably with image-jacobian.
Proposed system was simulated and realized with experimental level. And this thesis confirms the possibility of trajectory manageability with image jacobian through analysis of experimental results.
