Robot control using visual feedback has been one of the first applications computer vision researchers and engineers have thought of. Apparently the term visual servoing was used the first time by Gerald Agin in 1977 ("Servoing with visual feedback", Proc. of the Seventh International Symposium on Industrial Robots, Tokyo, October 19-21, 1977, also TN 149 Artificial Intelligence Center, SRI International) for a system composed of the SRI vision module and a robot manipulator which was able to locate small parts, grasp, and insert them. Today this system may appear naive for a number of reasons. The parts were essentially 2-dimensional, the images were coded with 1 bit per pixel and the camera was arranged such that the image plane be parallel with the table top on which the parts were lying. Even with these constraints, the computation power available two decades ago was not enough to process the images within the time limit imposed by the control loop of a manipulator. Nevertheless, this early work revealed the basic ingredients of a camera-based robot control system: robot control, camera geometry, and image-feature tracking.