Adaptive actuator failure compensation control for uncalibrated eye-to-hand manipulators

The compensation for unknown time-varying actuator failures is rarely considered in the existing visual control schemes for robotic manipulators, which can result in a degradation of transient performance in practice when actuator failures take place. Based on such a consideration, this paper presents a visual fault-tolerant control method for uncalibrated eye-to-hand manipulators with possible time-varying actuator faults. To accommodate the deleterious impacts of actuator failures on transient performance, firstly an adaptive control method is developed to compensate for time-varying failures occurring in actuators. Then, a pseudo inverse of image Jacobian matrix is designed to transform image error into joint space of robotic manipulators so that the unknown failures and camera parameters are successfully linearized, based on which a parametrization form of uncertain dynamics can be derived successfully. Moreover, an adaptive mechanism is newly presented to achieve the unknown camera parameters estimation online. With the aid of the Lyapunov method, the asymptotic convergence of image errors can be proved strictly. Simulation results are presented to demonstrate the proposed scheme.