Visual Servoing Technique On Nano-manupulator

Nano-technology is developing in a rapid speed.Due to the excellent properties of,nanomaterials are applied on various fields.Nano-manipulation techniques is playing a crucial role to the development nanotechnology.Conventionally nano operating systems are relying manually operating signal input devices to complete the operation of nanoscale material.This method is time consuming, and likely to cause damages to the equipment.As robot has highly successed in operating area, operation's automation in nanometers is a trend. Automatic nano-operation must implement in closed-loop control, and existing sensors have difficults to achieve nanometer level accuracy.Scanning electron microscopy(SEM),with its high resolution, high sampling rate becomes the indispensable tool for nano-operating systems.The paper raises that rely on scanning electron microscopy' image to achieve closed-loop control in nano-manipulation. Make use of visual servo technology to realize that driving a nano-probe to a certain point asigned by operator or some features in scene, we call that visual servo applied in nanometer scale.Summarized key technological in visual servo area and made some innovations.First, established a nano-manipulator systems and introduced the function of each part in the system.Selected SEM as a visual servo controller, briefly introduced the principle of imaging systems and settuped a simplified subjecting model of scanning electron microscope.Selected piezoelectric actuators as drive device of nano-probe to, and obtained a simplified model of the piezoelectric ceramic device;In order to achieve the rapid response speed and high acuracy, used a strain gauges coordinate with SEM to relize position feedback.While the system is largely depended on the SEM image, the performance of image processing should be optimized, as well as noise reduction and image drift compensation. The paper chose the NL-means denoising filter algorithm for that comparing to conventional noise reduction methods,it has more excellent performance, however this method needs large amount of computation,conventional computing methods can not achieve the real-time requirements of the servo system,we proposed an idea which relies on GPU parallel accelerated solution to the problem; Proposed feature image similarity algorithm for SEM image drift compensation; as for visual servo prerequisite,we must know the location of objects and controlled targets in real time,we used affine invariant SIFT algorithm to follow the movement of the probe tip. For tracking target,operators can use a mouse to choose the particular point or rely on auto-recoginze of intresting point by image processing methods.The goal of the project is to get the subject piezoelectric actuator input signal required by the SEM image.Through research the image-based visual servo control law in robotics, we migrate the laws to the nano-manipulator. Comparing the pinhole imaging model to the SEM imaging model,we adapt the image Jacobi to our system. Applied online adjustment algorithm for real-time graphics Jacobi matrix;For speed and accuracy of controlling nano-probe ,proposed switching method between visual servo signal and strain gauges signal; Proposed a simple and practical method of Z-axis contact detection.Finally, conducted a target tracking simulation on the image plane, simulation results are analysised and obtained optimal control parameters for PID controller; Used VRML software to build a three-dimensional visual servo control simulation platform and simulated the visual servo implementation process. Summarized the deficiencies and improve direction for micro-nano operating field with visual servo method.