Ultrasound-based visual servoing control of magnetic microrobot

Microrobots (MRs) hold the potential to revolutionize diagnosis and therapy. Thanks to their reduced dimensions, they have the ability to access and operate in hard-to-reach body districts performing submillimetric manipulation and targeted drug release. The safe operation of biomedical microrobots requires fine control capabilities, which strongly depends on precise and robust feedback about their position over time, and excellent movement abilities, essential to navigate in unstructured environments. In this thesis, a visual servoing platform for MR teleoperation has been developed. Magnetic actuation has been combined with ultrasound acoustic phase analysis (US-APA) motion tracking to achieve closed-loop navigation of a magnetic MR, rolling on the boundary of a lumen in a tissue-mimicking phantom. A robotic arm was used to position the magnetic actuation source and US-APA detection unit within the workspace, thanks to a C-arm system. The end user interacts with the robotic platform thanks to an intuitive graphical user interface (GUI), setting several imaging parameters, and to a joystick-based human machine interface (HMI), enabling robotic arm teleoperation. In the first place, the proposed approach allowed to perform supervised localization of the MR without any a-priori knowledge of its position. After localization, a robust real-time tracking enabled closed-loop MR teleoperation in the phantom lumina over a travel distance of 80 mm (145 body lengths), both in static and counter flow, thus achieving an average position tracking error of 368 micron (0.67 body lengths). These results validate US-APA as a reliable feedback strategy for visual-servoing control of MRs in simulated in-body environment