Uncalibrated Visual-Servoing of a Dual-Arm Robot for MIS Suturing

Abstract

In minimally invasive surgery small incisions are made in a patient's body. Long thin instruments and an imaging endoscope are inserted through the openings and a surgeon manipulates the tools using visual feedback from the endoscope. The difficulties of performing such surgery are outlined e.g. due to lack of dexterity the motions required can place tremendous strain on a surgeon's joints. This work looks at some of the issues involved in using visual-servoing to tie surgical knots with two robots. The constraints of the insertion points are managed through joint redundancy. The practical configuration of the robots is described. The effects of uncertainty of camera positioning and mounting of the surgical instrumentation by busy medical professionals, who are interested in clinical utility and whether they can see the operating area, are removed through uncalibrated servoing and a kinematic model of the robots. A novel encoder on the tools is used to determine the angle of tool orientation which is critical for aligning the jaws of the tool with the thread and body structures. A visual Jacobian is used to create the desired motions and control the redundant joints. A partial knot-tying algorithm is introduced which is based on an analysis of manual knot tying. Practical results of simulated intracorporeal knot tying are then given