Vision-based Automated Manipulation for Magnetic Microrobots: from Individual to Swarm

基於視覺反饋的磁性微型機器人的自動化操作:從單體到集群

Student thesis: Doctoral Thesis

View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date2 May 2023

Abstract

The in vivo manipulation of magnetic microrobots and microparticles has attracted considerable attention because of its advantages of noninvasiveness and high precision in the targeted delivery in the biomedical applications. However, the automatic manipulation of microrobots in the dynamic environments with fluid flow remains challenging due to the small size of these robots, the unknown dynamics and disturbances caused by the fluid flow, and the lack of an effective real-time in vivo imaging thechnique for locating them. Moreover, the drug-carrying capacity of a single microrobot is inevitably restricted to its size. Thus, swarm microrobots which contain numerous magnetic microrobots or microparticles have shown greater potential than a single microrobot because of higher carrying capacity and easier-to-track property (by in vivo imaging systems). In this thesis, an automated control scheme is proposed for a magnetic micro-rotor (an anchor-like microrobot) navigating in the dynamic environments, and an automated control strategy is presented for magnetic micromanipulation of microswarms in a rotating gradient field. The thesis includes two major parts.

In the first part, a micro-rotor is designed to be levitated to achieve a holonomic movement in the magnetic field under clinical imaging. The micro-rotor is a type of microrobot that has the potential for microsurgery and delivery. It is also used as an effective example of microrobot in navigation control. Then, a front-end optimal path planner that considers both the path length and the path clearance based on the particle swarm optimization, as well as a back-end dense corridor-based trajectory generator, is proposed to ensure the collision avoidance during micro-rotor navigation in blood vessels. That is, a collision-free trajectory is designed to navigate the micro-rotor to the desired position accurately while hovering it in the near center region of the blood vessel so as to achieve decent trajectory clearance and optimal control effort. Here, the extravascular environment is regarded as static obstacles. To achieve robust and safe trajectory tracking of the micro-rotor in dynamic environments, a model-predictive controller based on the extended state observer is used to compensate for the unmodeled dynamics and unknown disturbances while further restricting the state space. Unlike traditional approaches that only impose simple constraints on the robot state, in the proposed controller, corridors are used to restrict the micro-rotor state in the control, and then check the safety of the system output. Simulations are performed to tune the parameters and validate the proposed approach. Experiments of navigating the micro-rotor in dynamic environments with different flow rates both in 2D and 3D spaces are carried out to demonstrate the effectiveness of the proposed approach.

In the second part, an automated control scheme for micromanipulation of microswarms in a rotating gradient field is proposed. Different from the rotating uniform magnetic field generated by Helmholtz coils, the convergent field, which is established based on the rotational gradient magnetic field, can drive the microswarm to a determined position and facilitate the swarm localization and tracking. According to the preliminary motion characterization, an intuitive trapping dynamic model that does not require complex force analysis is established. On the basis of this dynamic model, a robust controller with disturbance compensation is designed to manipulate swarm microparticles to track a desired trajectory. A second-order sliding-mode estimator based on the super-twisting algorithm is designed to estimate the position of the microswarm and disturbances. The proposed control scheme under the convergent field can avoid volume loss and unexpected drug diffusion of the swarm when facing complex in vivo environments. The stability of the control scheme is proven by the Lyapunov approach. Experiments are finally performed to demonstrate its effectiveness.

In summary, this study proposes feasible automated manipulation schemes for microrobots and microswarms that help them precisely navigate in microenvironments. It provides a meaningful reference for the future biomedical applications of various microrobots and microparticles.