Abstract
Despite advances in automated biological cell micromanipulation, most micromanipulation systems rely heavily on calibration effort to ensure functionality on single cell operation. With increasing demand for quick response and in-situ cellular investigation, laboratory-based micromanipulation system is falling short due to its lack of mobility as well as tedious preparation work. In order to enable on-the-fly deployment, portable micromanipulation has been proposed and developed using existing computer vision methods. To address the issue for quick deployment, this thesis proposes a novel lowcost vision-based micromanipulation system on a portable setup with minimal calibration. Various technical challenges in a minimally calibrated vision-based micromanipulation are addressed, including detection, navigation, and interaction of the end effector and cell specimen. The methods proposed in solving the technical challenges have been demonstrated on plant cell as multi-cell specimen, using a built in-house portable micromanipulation system. Qualitative and quantitative results are presented together with discussion for potential future directions in formulating and applying a vision-based portable micromanipulation system. This thesis fills a gap in the literature by proposing methods that address the dynamic environment faced in a less than ideal micromanipulation, without the needs of extensive calibration or intricate equipments.