Vision based platform for drug delivery microrobot

Abstract

In recent decades, permanent magnet end-effectors have been used in several fields and applications due to their unique ability to remotely control any magnetized object’s positions via delivering the magnetic flux through space and non-magnetic materials. The reason of choice for magnetic end-effectors is because of their precision of control in comparison to other analog systems. This end-effector is able to move in three dimensions with a micro-precision of 0.05mm. In this thesis, this unique ability of magnetic material is further advanced to produce an optimized local maximum magnetic field. As part of this work, a Genetic Algorithm and Exhaustive search algorithm are implemented to get this optimized magnetic field. This method is suitable for medical domain applications such as magnetic microrobot drug delivery. A 3D actuator system had been designed to control the magnetic end-effector for this purpose. An open-loop system architecture is proposed to control the movement of the microrobot. The results of the microrobot drug delivery experiment proved to be promising. The results show that the end-effector is able to navigate the microrobot in 2-dimension space, having an error of less than 0.1mm. A tracking system based on the Lucas-Kanade Optical Flow algorithm is implemented using the open-CV library. The optimized magnetic field was validated in an experiment using a 3D Tesla meter.