Data-Driven Remote Center of Cyclic Motion (RC$^{2}$M) Control for Redundant Robots With Rod-Shaped End-Effector

Abstract

Remote center of motion (RCM) has become a rising research direction in the field of robotics. It means that a robot with a rod-shaped end-effector operates through a tiny hole on the surface. Thereinto, a key issue is the deviation of the RCM point's position, apart from the operating accuracy. In addition, considering that an RCM robotic system generally consists of a commercial robot and a specialized rod-shaped end-effector, there exist some errors in the structural information related to the attached end-effector. In this article, a remote center of cyclic motion scheme with a data-driven technology is proposed to control robots, of which end-effectors' structural parameters are inaccurate. Meanwhile, a recurrent neural network is proposed to figure out the scheme's solution, with the relevant theoretical analysis given. Furthermore, simulative and physical experiments on a FRANKA Panda robot with a rod-shaped end-effector are conducted to validate the control scheme's effectiveness distinctly.