Session: 16-02-01: Poster Session: NSF Research Experience for Undergraduates (REU), NSF Posters

Paper Number: 99721

99721 - Integration of Autofocus for Real-Time Monitoring of High-Precision Roll-to-Roll Printing Process 

High precision production quality control is highly dependent on the inline inspection of manufacturing processes, e.g., the real-time monitoring and metrology for the microcontact printing of flexible electronics. So long as image focusing is required, these image techniques can cause manufacturing bottlenecks due to their slower autofocus (AF) methods. The measurement of printed pattern production can be aided by the invention of many high-speed and high-accuracy image acquisition and image processing techniques. However  many of these techniques are not sufficient for real-time industrial inspection due to their lack of integration into real-time applications. In this project, we propose to integrate a fast autofocus algorithm into a piezoelectric motion stage (PEMS) for real-time inspection of a high-precision roll-to-roll printing process for flexible electronics. As the printed pattern in roll-to-roll has micro-scale precision and moves on a flexible substrate, fast-speed autofocus by integrating software and hardware is demanded. We propose that using the piezoelectric motion stage (PEMS) for position control would significantly decrease the AF time, bringing the frequency of operation from 2 Hz to 100 Hz. Instead of using an internal controller to account for the problematic hysteresis effects of the piezoelectric motion stage, we use the long short-term memory unit to integrate the hysteresis effects and the focus measurement into a single learning-based model. Subsequently, based on this long short-term memory model that successfully finds the optimal focus position using a series of focus measurements derived from a sequence of images, a model predictive control method is developed. The PEMS-based AF system is assembled and tested on a roll-to-roll system. The assembly consists of the camera and tube lens attached above the PEMS, which is attached above the manual motion stage. We will attach the PEMS-based assembly at an inclined angle to the frame of the roll-to-roll system and then test the AF algorithm. The camera can capture the pattern on the roller, while we monitor the AF speed and accuracy. The heavy weight of the camera and the long length of the tube lens causes jitters in AF. A sufficient way to solve jitters is to increase the AF time to allow time for the s-curve to reduce the jitters as much as possible. Our proposal has proven to be in line with our experimental results which showed us that it takes 6 images for the AF process to get the focused image of the moving objects which is about 150 ms, reducing the AF time by a minimum of 30% when compared to well-known ruled-based AF methods and other learning-based methods. 

Presenting Author: Ahmed Elbashir UMass Amherst

Presenting Author Biography: Ahmed Elbashir is working towards his B.S in mechanical engineering at the University of Massachusetts Amherst.

Authors:

Ahmed Elbashir UMass Amherst
Jingyang Yan UMass Amherst
Xian Du UMass Amherst