Design, Sensing, and Control of Soft and Continuum Robots with Application to Medicine

Abstract

The overarching goal of this dissertation is to advance the capabilities of continuously flexible robots through new design, sensing, and control strategies, with a focus on medical applications. Continuously flexible robots are one of the major frontiers in the robotics research due to their high dexterity and flexibility. While their design and actuation has received considerable attention, accuracy and precision in using them remains a challenge due to limitations in sensing and control strategies. There is a clear need for better sensing methods with higher resolution, continuous if possible, to enable precise manipulation for these robots. In addition to those, more advanced control methods that consider the compliance and the uncertainty around the shape of the robot are needed to make full use of their potential. This dissertation aims to address these needs. In Chapter 2, a novel shape sensing method for soft robots is introduced. The method uses Electrical Time Domain Reflectometry to detect the continuous shape of the robot via impedance changes along embedded wires. Chapter 3 focuses on an Extended Kalman filter that infers the missing degree of freedom in magnetically tracked needle pose by explicitly modeling torsional windup, enabling closed-loop control of torsionally flexible needles. Using this approach, needle steers in phantoms and lungs were performed with good accuracy, a major step toward future lung cancer treatments. Chapter 4 applies hybrid force-motion control to a parallel arrangement of coupled flexible needles operating in free space. This controller is based on a Cosserat rods model, and accounts for robot kinematics and deflections under load. We show that this approach is feasible for force regulated path following with parallel continuum robots.