High-speed 3D optical sensing for manufacturing research and industrial sensing applications

Abstract

This paper presents examples of high-speed 3D optical sensing for research and applications in the manufacturing community. Specifically, this paper will focus on the fringe projection technique as a special technology that can be extremely beneficial to manufacturing applications, given its merits of simultaneous high-speed and high-accuracy 3D surface measurements. This paper will introduce the basic principles of 3D optical sensing based on the fringe projection technique as well as the enabled manufacturing research applications, including both in-situ/in-process monitoring and post-process quality assurance.

Corrosion characterization of engine connecting rods using fringe projection profilometry and unsupervised machine learning

“…Each of these metrics is a measure of different aspects of surface roughness, as mentioned above. The accuracy of using this acquisition method in comparison with other methods to measure roughness is validated in [26,33]. Additionally, the comparison between the actual surface morphology and morphology obtained from reconstruction is illustrated in section 4.…”

“…(vii) Kurtosis (S ku ): he fourth central moment of the amplitude probability function that represents the sharpness of the profile is called Kurtosis coefficient. [26,35,36] The measurement of roughness metrics using structured light has been performed by Windecker et al [37]. However, the system is highly complex and very large in size.…”

“…As mentioned above, the use of neural networks possesses some disadvantages, like lack of interpretability and dependence on the volume and quality of the training data for generalization. In order to avoid the use of neural networks to classify images directly, we propose to use the roughness parameters that have been determined by the method described by Li [26] and then classify them into 3 corrosion levels based on the calculated parameters, using an unsupervised ML algorithm, namely K-means clustering. An outline of the project pipeline has been given below in figure 1.…”

Recent Developments

“…A Feature Paper can be an original research article or a comprehensive review of the latest advances in the field. In this issue, we have two such pieces: "Applications of Digital Twins in Power Systems: A Perspective" by Leila Kamyabi, Tek Tjing Lie, and Samaneh Madanian [3], and "High-speed 3D optical sensing for manufacturing research and industrial sensing applications" by Beiwen Li [4].…”