Reduced errors in robot-aided minimally invasive surgery through online condition number optimization

Abstract

The use of robots in surgery is of great research interest following the widespread adoption of da Vinci systems in surgical rooms worldwide. Research interest has grown, as such robot aid still cannot fully perform procedures requiring smaller instruments and more precise movements. In addition to added flexibility, serial-link manipulators can also provide these capabilities for surgical procedures. In surgical procedures, robot control techniques should strive for low errors along with slow joint speeds. This paper discusses control theory and conducts an analysis on the shortcomings of damped inverses and singular value decomposition inverses. The use of online local optimization of the condition number of the Jacobian to increase the dexterity of redundant serial-link manipulators with arbitrary geometries was proposed. Techniques in previous studies were compared in a simulated minimally invasive surgical task. The proposed technique utilized inner motions to autonomously reposition the manipulator in high dexterity configurations without disturbing the main task. Hence, the proposed technique decreased the remote center-of-motion error by 27% and reduced the joint velocity requirements for the same trajectory by 34%.