Author: Mohammad SalehizadehPublisher:
ISBN:
Category :
Languages : en
Pages : 0
Book Description
My PhD dissertation takes an innovative approach to using the fundamental tools of robotics and control to solve the underactuated control problem of multiple magnetic microrobots for biomedical applications. The ability to use a team of microrobots to run a task can offer many advantages. However, the integration of on-board powering and sensing circuitry has not yet become possible at the microscale. Therefore, all magnetic microrobots (micro-agents) of the team have to share a single driving signal, whereas the system has multiple states to be independently controlled. Another major challenge with magnetic team control is that when multiple magnetic microrobots work together in close proximity, the agents tend to irreversibly stick together due to strong magnetic inter-agent forces. Previous studies either ignored the inter-agent forces by simply assuming their robots were far apart or treated these forces as disturbances without verifying the stability in close proximity. I solved for the first time this problem for a pair of magnetic agents in close proximity in 2D (and later in 3D) by making full use of inter-agent forces to control the motion of agents. In my approach, the positions of microrobots were controlled independently. As a practical demonstration, I showed for the first time that the motion of two functional magnetic microrobots as microgrippers can be controlled in 3D to run a task. Subsequently, to generalize the inter-agent force control to more agents, I introduced two solutions: 1) via optimization-based control (as a control technique), and 2) via motion planning and tracking (as a robotics technique). My PhD research enabled for the first time the collision-free autonomous navigation of a team of magnetic microrobots in close proximity. This solution allows magnetic microrobots to potentially act in a cluttered environment, such as the human body or microfluidic channels. To this end, I employed rapidly-exploring random tree (RRT) motion planning. I further incorporated the idea to run two practical demos that could be applied to cell testing/manipulation. The results of this PhD work can be applied to actuation and sensing, especially in the design of field-activated medical devices and for localized targeted drug delivery.
