UKF-Based Motion Estimation of Cable-Driven Forceps for Robot-Assisted Surgical System

This paper presents an Unscented Kalman Filter (UKF)-based method to achieve high-precision motion estimation of cable-driven forceps for a robot-assisted surgical system. We analyze the operational/working principle of revolute joints of a 3-degree-of-freedom (3-DOF) cable-driven surgical manipulator. Then a gripper jaw is selected as a representative joint, which is actuated by a single-motor cable-driven mechanism with a reset spring. The corresponding system dynamics comprehend the mass, elasticity, damping, and friction of steel cables. By using the displacement and velocity of reset cable and the rotation angle of motor as observations, the motion estimation model based on UKF is derived. The estimation accuracy is verified experimentally, with the errors of absolute and root-mean-square (RMS) of less than 0.5 deg and 0.2 deg respectively. By comparisons with the least square methods (LSMs), the installation strategy of only one displacement sensor on the reset cable is determined, which is conducive to further refinements of the mechanism. Furthermore, the external force loading experiments are performed, with the RMS estimation error of less than 0.5 deg for the external force of no more than 250 g applied on the tip of the gripper jaw. These experimental results validate the motion estimation accuracy of cable-driven forceps, without requiring sensors on the end joints or slender tool shaft of surgical instruments.