Recently, CCD camera is widely used for vision based automation system which works successfully in unstructured environments.
In this thesis, a manipulation system with a CCD camera is implemented by two approaches: a calibration-based approach and a visual servoing approach. The former is a well known and classical method, the latter is a popular control strategy that precisely positions a manipulator or an object using imprecisely calibrated camera-lens-manipulator system. The calibration-based approach is achieved by two calibrations, one for computing the camera intirinsic parameters, and the other for the transformation between the camera and the manipulator. The visual servoing approach is achieved by combining a jacobian estimator and a propositional velocity controller.
Two approaches are tested for a coil loading system to identify difficulties in real implementation and error sources. And then, we evaluate the feasibility of two approaches for an automation system based on the comparison of their merits and demerits.
