Estimation method for normalised feature Jacobian matrix using only one camera

For feature-based visual servoing of an eye-in-hand robot, it is necessary to know the relationship between small feature changes and differential motion of the camera mounted on the robot´s end-effector. This paper proposes a new scheme for online estimation of this relationship. This proposed scheme uses only one image and requires no a priori knowledge about environment, and therefore would be an effective tool for the feature-based visual servoing.