Design and Mechanics of Continuum Robots for Surgery
AUTHORS
ABSTRACT
This dissertation describes design and modeling of two new flexible, continuous backbone robotic dexterity enhancement devices, and algorithms to robotically manipulate them in surgery. These steerable needles and active cannulas provide dexterity in thin (needle-sized) form factors, permitting surgical tools to “turn corners” inside the human body. The robotic tool manipulation algorithms are useful for rapidly and accurately aligning many kinds of surgical tools with planned poses and entry vectors. Drawing on techniques from artificial intelligence, they do not require cumbersome calibration, complete knowledge of robot kinematics, or even encoding, making them well-suited to uncertain real-world clinical environments. Simulations and experimental results demonstrate the accuracy and speed of the algorithms. When an intervention is needle-based, steerable needles provide a means to achieve dexterity within soft tissue after insertion begins. Asymmetric forces generated by a bevel tip are harnessed robotically to generate controllable deflection, and curvature is enhanced by design of needle properties and geometry. The needle is modeled as a nonholonomic system with model parameters fit to experimental data. A new control law inspired by mouse ballistics enables human users to teleoperatively target more accurately than traditional rate or position control. Active cannulas are capable of similar curved dexterity without relying on tissue reaction force, making steering possible in both soft tissue and free space. Composed of multiple concentric precurved tubes, active cannula shape is described using Bernoulli-Euler beam mechanics and minimum energy principles, and predictions are corroborated with a set of experiments. Parameter values fit to experimental data are near ranges calculated from physical tube characteristics. Design tools for surgical application-specific cannula optimization are described, and differential kinematics are derived, enabling future work in image-guided control and teleoperation. The research in this dissertation has been motivated by reducing invasiveness, improving clinical outcomes, and enabling surgical treatment for “inoperable” patients. However, the models and methods developed are broadly applicable within the fields of continuum and nonholonomic robotics.
Tags: dissertations