Adaptive task space PD control via implicit use of visual information

Visual feedback is very useful when the precise kinematics of the manipulator are not available. However, visual information may be lost during manipulation, which may make the system unstable. This paper proposes an adaptive task space PD feedback control method for setpoint positioning of robotic manipulators that is robust in the case of a lack of visual information and kinematic uncertainty. The key idea of this approach is that visual information is not used directly in the feedback loop, but only for adaptation of kinematic parameters. This paper describes the design of a controller and parameter update laws, and proves the stability of the system with and without visual information. Selected simulations are executed to validate the proposed controller.