Stochastic Approach for Feature-Based Tip Localization and Planning in Nanomanipulations

In atomic force microscopy (AFM)-based nanomanipulation, the tip position uncertainties still exist due to the parameter inaccuracies in the open-loop compensation of the piezo scanner, the noise in the closed-loop control and thermal drift. These spatial uncertainties are very challenging to be directly estimated owing to the lack of real-time feedback, and its effects are more significant in performing an automatic nanomanipulation/assembly task than macro world manipulations. In this paper, we propose a stochastic framework for feature-based localization and planning in nanomanipulations to cope with these uncertainties. In the proposed framework, some features in the sample surface are identified to calculate their positions in statistics, and detected by using the AFM tip as the sensor itself through a local scan-based motion. In the localization, the Kalman filter is used through incorporating the tip motion model and the local scan-based observation model to estimate the on-line tip position in the task space. The simulation and experiments about tip positioning are carried out to illustrate the validity and feasibility of the proposed algorithm. Then, positioning tip for effective nanomanipulation is presented by using several experiments. Finally, a carbon nanotube is followed to show that the proposed method can provide a great potential for improving the position accuracy.