Visual Servoing Control Based on Reconstructed 3D Features

Image-based visual servoing (IBVS) is an automatic control technique which uses image information to control the camera moving to a desired location. To solve the problems of unsmooth trajectories and slow convergence rate caused by classical IBVS during the servo process, a visual servoing method based on reconstructed three-dimensional features is proposed. First, the camera pose was estimated, to obtain the depth information of the target by solving the Perspective-Four-Point (P4P) problem. Then, the new three-dimensional features for visual servoing is constructed. Finally, the visual servoing model is established based on classical IBVS control structure in terms of cuboid. Simulation results show that the proposes method has higher convergence speed and smoother convergence trajectories compared with the classical IBVS method.