Continuous guidance for the docking of autonomous vehicles using active task-space sensing

In this paper, an active line-of-sight-sensing (LOS) methodology is proposed for the docking of autonomous vehicles/robotic end-effectors. The novelty of the overall system lies in its applicability to cases that do not allow for the direct proximity measurement of the vehicle´s pose (position and orientation). After the long-range motion of the vehicle, the proposed method first determines the accumulated errors in localization and, then, continuously guides the vehicle, via active sensing, while reducing these errors below the random noise level. Multiple LOSs are utilized in defining a continuous, gradient-descent trajectory for the vehicle motion. The proposed system was successfully tested via simulations and experiments on a 6-dof (degree-of-freedom) vehicle and 3-dof high-precision platform.