Uncalibrated dynamic eye-in-hand visual tracking of robots by observing unknown static objects

Uncalibrated visual servoing techniques based on the depth-independent interaction matrix framework have been proposed for the general 3D motion control of robot manipulators, which can simultaneously handle both the unknown intrinsic and extrinsic camera parameters. By extending the existing adaptive eye-in-hand visual servoing approach based on the depth-independent interaction matrix, in this paper, we propose an uncalibrated dynamic visual tracking approach for the eye-in-hand camera configuration by observing multiple feature points of a static object. The proposed approach can simultaneously handle the unknown camera parameters and the unknown 3D coordinates of the feature points, where a new adaptive law is developed to estimate these unknown parameters online. On the basis of nonlinear dynamics of robot manipulators, we also prove the asymptotic convergence of the image errors with the help of Lyapunov theory. To demonstrate the performance of the proposed approach, preliminary simulation results will be given by using a two-link planar robot manipulator.