Research

Motivation: Intracerebral hemorrhage (ICH), which is one of the most common forms of stroke, occurs when blood from a ruptured vessel accumulates within the cranial cavity. While continuum robots have been proposed to enhance hematoma coverage/evacuation, the access trajectory from the burr hole to the hematoma remains straight. We intend to explore new continuum robot design that is capable of simultaneous nonlinear navigation in the brain and highly decoupled, dexterous tip manipulation.

Main work: A hybrid continuum robot, consisting of a precurved body and a 2-degree-of-freedom (DoF) flexible tip monolithically fabricated, has been developed to enable simultaneous nonlinear trajectory following and dexterous distal manipulation for safer intracranial access and more complete surgical task execution. Dedicated design optimization and motion planning algorithms are developed to enable the 2-dimensional follow-the-leader (FTL) motion of the cannula. Simulated results show the ability of the cannula to travel along a preplanned curved trajectory with minimal shape deviation. The motion capability of the robotic cannula was further experimentally proven in a brain-like gelatin phantom.

Motivation: Model-based control of continuum robots often suffers from inaccurate modeling of the flexible behavior due to their structural deformation, interaction with soft tissue, and collision with other instruments. We strive to contribute to model-free and data-driven control of continuum robots to enable robot manipulation with high accuracy and computational efficiency for safe application in surgery.

Main work: A model-free multiagent reinforcement learning (RL), namely multiagent deep Q network (MADQN), has been introduced to control a cable-driven continuum surgical robot. The control of the 2-degree of freedom (DoF) robot is formulated as a one-DoF, one agent problem in the MADQN framework to improve the learning efficiency. Combined with a shielding scheme that enables dynamic variation of the action set boundary, MADQN leads to efficient and safer control of the robot. Shielded MADQN enabled the robot to perform point and trajectory tracking with submillimeter root mean square errors under unmodeled external loads and soft collision. 

Motivation: Magnetic resonance imaging (MRI) offers high soft tissue contrast and thermometry capability, making it an ideal imaging modality for intraoperative robot-assisted brain lesion ablation.

Main work: A 2-degree of freedom (DoF) magnetic resonance imaging (MRI)-conditional continuum robot has been developed with two pairs of high- performance modular shape memory alloy (SMA) spring actuators and a 3D-printed flexible end effector. The robot is capable of large bending motion in 3D workspace. Robot kinematics and SMA model have been developed to enable the SMA actuation configuration initialization and improve robot control. Preliminary tests in the MR scanner room show that the robot under continuous operation results in MR images with negligible signal-to-noise ratio (SNR) reduction.  

Motivation: Ultrasound (US) imaging is commonly used in percutaneous needle procedures for real-time guidance. Accurate and robust automatic needle tracking is highly desirable in both manual and robotic needle manipulation due to the relatively low imaging quality of US, dramatic target appearance change, presence of distractors, and issue of temporary target disappearance. 

Main work: We have introduced various learning approaches and filtering strategies to advance the capability of existing needle tracking algorithms, including (1) improved compressive tracking (ICT) algorithm and a modified Sage-Husa adaptive Kalman filter (SHAKF) and (2) fusion between visual tracking module and motion filtering module. Our latest tracking systems have shown distinct improvement against other state-of-the-art trackers during the motorized and manual needle insertion experiments in both gelatin phantom and biological tissue environments. 

Motivation: Transoral robotic surgery in upper aerodigestive tract remains challenging due to the rigidity and insufficient workspace of existing instruments, and inappropriate balance between compliance and stiffness in flexible instruments. 

Main work: A flexible surgical robotic instrument, with four degrees of freedom (DOFs), featuring a shape memory alloy (SMA)-based continuously variable length bending joint (CVLBJ), was developed for TORS. The integration of the CVLBJ allows bending length adjustment, leading to enhanced distal workspace compared to traditional fixed length flexible instruments. The instrument also exhibits variable stiffness capability via the length variation of the CVLBJ and the SMA temperature variation.

Motivation: There exist topological obstructions to achieve robust and global asymptotic stability results for any systems involving rotational motion. This could introduce operating restrictions and affect the robustness in surgical robot control. 

Main work: To overcome the topological obstructions, we propose synergistic hybrid feedback algorithm for robustly and globally asymptotically stabilizing/estimating a particular attitude. The main idea is to design a family of potential functions with the synergism property (i.e. there exists a decreasing potential function at all undesired critical points). Then, robust and global asymptotic stability results can be guaranteed through duly selecting the gradient-descent state feedback in conjunction with a hybrid switching mechanism. In comparison with existing hybrid feedbacks in the literature, our results provide more complete solutions with the central synergism property.