RobPosit, a robust pose estimator for operator controlled nanomanipulation

Abstract

Nanomanipulation is an important approach for recent developments in nanoscale production, characterization and analysis, both in material and life sciences. Its use is actually limited because of its complexity. Especially inside a scanning electron microscope, kinematic constraints, open-loop actuators and lack of natural visual access are barriers for its widespread use and require an operator specialized on a given set-up. We propose a novel algorithm that can robustly and automatically track the 3D pose (i.e 6 DOF) of a marker placed on the frame of the SEM. This algorithm does not need operator intervention. The algorithm is coupled with a virtual model of a manipulation set-up, synchronized in real-time with the real set-up, for facilitating grasping and control of the operator during the manipulation task. Moreover, such an approach would provide a virtual feedback to implement an overall closed-loop control on the set-up.