Visual Servoing of manipulator with moving base

Manipulator systems are necessary part for unmanned autonomous systems (UAS) to achieve task executions and hence, they have been widely adopted in mobile platforms, such as UAV, UGV together with visual feedback to perform different tasks. Due to the unknown movements of the base, it is difficult to achieve the accurate manipulator visual servoing control since the depth information is continuously changed. In order to achieve the manipulator visual servoing with moving base, two solutions are presented in this paper. Firstly, a support vector regression based method is developed to predict the horizontal movements of the base. Thus, the position of the end-effector can be adjusted accordingly so as to reduce the influence of moving base on the end-effector. Then, an observer is proposed to estimate the depth change in presence of disturbance acted on the camera velocity. With the estimated depth, a visual servoing controller is developed. Simulation and experimental results are presented to illustrate the performance of the proposed solutions.