MODELING AND TASK AUTOMATION FOR FLEXIBLE SURGICAL MANIPULATORS VIA DATA-DRIVEN APPROACHES

Abstract

With the advancement of new surgical paradigms such as keyhole surgery and natural orifice transluminal surgery, flexible/soft robot manipulators have become promising solutions due to the capability of navigating through torturous and confined tissue environments and reaching surgical sites that are inaccessible to their rigid counterparts. Their inherent compliance brings about safer tool-tissue interaction. The unique actuation mechanism such as tendon-driven method enables the separation of distal manipulators from the bulky proximal actuation systems. Nevertheless, challenges exist in modeling and controlling these robots due to actuation redundancy and various nonlinear effects. The level of complexity increased when unknown tool-tissue contact dynamics are involved. Data-driven approaches provide an efficient tool to model the kinematics and derive control policies without prior knowledge. Therefore, my thesis aims at leveraging data-driven methods to address the above issues and ultimately achieve the automation of surgical sub tasks for flexible/soft surgical manipulators.