A dynamic active constraints approach for hands-on robotic surgery

Abstract

Toward the goal of developing a hands-on robotic surgery control strategy which simultaneously utilizes the various strengths of both the surgeon and robot, we present a dynamic active constraint approach tailored for hands-on surgery. Forbidden region active constraints are used to prevent motion into areas which have been deemed dangerous by the surgeon, helping to overcome some of the disadvantages of fully active systems such as loss of tactile feedback, limited workspace, and limited field-of-view. The computer graphics technique of metaballs is used to represent point cloud data from an imaging system with an analytical, differentiable surface and a dynamics-based controller is proposed which controls the robot to lie on the zero set of the generated time-varying implicit function for which the motion is either known or unknown. This controller has been incorporated into a recursive null-space approach to allow for unimpeded motion along the surface and for further extension to joint optimization in the future. This methodology is demonstrated in simulation and on a lightweight, seven-degree-of-freedom serial manipulator.