Uncalibrated visual-servoing of a dual-arm robot for surgical tasks

In minimally invasive surgery, small incisions are made in a patient's body and tools are inserted through these incisions. A surgeon performs operations using visual feedback from camera images. Such procedures can be very strenuous to perform. Various steps have been taken to utilize robots in minimally invasive surgery. The vast majority of such systems are teleoperated, with very little work on autonomous systems. This work looks at the potential of using uncalibrated visual-servoing robotic systems as an aid for surgeons. The range of tasks involved in minimally invasive surgery present a good set of tasks for the further development of visual-servoing techniques. In this paper we present a testbed that is used to develop visual servo algorithms for the completion of complex surgical tasks. A sample task involving manipulation of a thread using the two grippers is evaluated for its implementation efficacy.