Khamesee, Mir BehradNofech, Joseph2024-12-172024-12-172024-12-172024-12-11https://hdl.handle.net/10012/21264This study presents a novel methodology for achieving three-degree-of-freedom (3-DoF) control for an attractive-type magnetically-levitated (maglev) microrobot using machine learning. Traditional micromanipulation methods face challenges associated with friction and maintenance requirements; particularly in applications such as cell injection, where current devices tend to be limited by their high cost and maintenance requirements. The precise and low-maintenance nature of attractive-type levitation makes it a viable alternative to traditional micromanipulation methods, but a primary challenge lies in the difficulty of achieving precise 3-DoF control for such systems due to the complexity in the magnetic fields they generate. This research addresses this challenge by introducing a machine learning-based methodology that automates the learning of levitation dynamics across the workspace. Our presented approach introduces and incorporates an automated system for generating training data with minimal human intervention, enabling a machine learning model to learn how the levitated microrobot responds to system inputs. This information is then used to establish 3-DoF position control of the levitated microrobot. Our automated methodology simplifies the setup process for new and newly-modified attractive-type levitation platforms, and is demonstrated to improve performance over conventional methods by accounting for observed variations in the levitation dynamics throughout the workspace; achieving up to a 20% reduction in root mean square error during trajectory tracking and up to a 36% reduction in step response settling times. The results demonstrate the ability of our automated methodology to significantly reduce the accessibility barriers associated with establishing and modifying attractive-type maglev platforms; effectively replacing the usual methods of finite element simulation, precise magnetic field measurements, and/or analytical calculations while providing enhanced levitation control over traditional methods. This advancement contributes to the field of micromanipulation and microforce sensing by offering a more accessible and efficient approach to achieving precise control in attractive-type maglev systems.enmaglevmagnetic levitationmachine learningmagnetismmicrorobotsmicromanipulationneural networksautomationMaster Thesis