Visual Servoing Decoupling Control Based on Line Features and Inner-region Based Features

To solve the problem for a multi-degree-of-freedom manipulator approaching an arbitrary quadrilateral target, we propose a method of combining line features and inner-region based features to achieve a robotic decoupled visual servoing (VS)process. We build the inner-region based features to guide the camera's translational velocities, while using the target's line features to generate the camera's rotational angular velocities, and then introduce a vector compensation for the inner-region based features and a position compensation for the centroid coordinates, resulting in a partially decoupling of the camera translational and rotational control. Finally, we carry out the stability analysis of the robotic VS control system. Simulation results show that the propose method can control the camera to converge to the desired pose with faster and smoother motion, and can better overcome the uncertainty caused by depth estimation under the condition that the camera optical axis is approximately perpendicular to the target plane.