(Source: Pongsakorn.M/Shutterstock.com)
2D-machine vision has transformed automated manufacturing, applying elegant software algorithms to the production, inspection, and control of mass-produced parts. This technology uses imaging to build a holistic, 360-degree view of a part and process to consider all factors that go into fabricating the part. The integration of software and advanced industrial equipment has driven physical human contact between the machine and the manufactured part.
Most beneficial for repeated, high-quantity production runs, machine vision provides the operator enhanced visibility to improve the quality, speed, and cost of production. 2D-machine vision offers enhanced inspection by detecting the position of product features, improving the defect rate for quality assurance. It increases productivity and flexibility by rapidly conducting repetitive movements and enabling pre-programmed changeover to batch a sequence of production steps, similar to subroutines in computer programming. Finally, 2D-machine vision reduces cost by these improvements to quality and speed. The improved programmability optimizes machine performance and reduces the human assets needed to run the machines. Better quality also translates to lower scrap, an immediate improvement to cost.
With all its disruptive benefits, 2D-machine learning carries some outstanding opportunities for improvement. The machine creates the target image using light. As a result, this approach is susceptible to variations in lighting conditions, such as shadows, shading at different times of day, thus impacting the crispness of the image. Because it is planar, the 2D approach is mainly suited for binary assessment, such as whether a feature is present or whether a defect is present.
These challenges create the need for an even further improved imaging solution: 3D. Simple in concept, adding depth to the 2D image requires significant upgrades to the entire process. Here we will outline how the third dimension extends the dynamic response and process performance gains of 2D-machine vision, moving the process ever closer to the ideal state: zero-defect, just-in-time, lowest cost.
The added complexity of incorporating the third dimension increases the computational load on the processors exponentially. To address this, software providers have improved their infrastructure, resiliency, and on-demand computation support. In parallel, 5G’s expansion helps to mitigate the processing constraint. It is easy to see why adding the third dimension would improve the imaging approach at the cost of added processing because higher amounts of data take longer to process. But sensors capturing a feature’s three-dimensional view enables the software to interpolate through an imperfection in the image to create an accurate picture using the feature’s position in the other two dimensions. Imaging process resiliency increases with a more detailed view, reducing the machine’s response time.
The software and algorithms collect and analyze data, accelerating the response to a failure without the need for human intervention. The process removes significant sources of human error and reduces the time between signal and response with a more accurate picture. Reducing the response time moves the process closer to the ideal state pillar of just-in-time production.
3D-machine vision enables a faster, more accurate view of the relevant feature. This benefit improves performance by finding and addressing quality failures. This view allows the engineers to define or pre-program a procedure for assessing the severity of a part integrity issue. They can optimize the machine to look at the known sources of error to improve operating efficiency, decreasing part cost.
Adding the depth dimension can improve the accuracy of measuring and cutting, enabling a tighter manufacturing tolerance of a process. Achieving a repeatable, tight tolerance could allow a more automated degree of manufacturing, which decreases the per-piece production cost.
3D-machine vision can compare a finished part against its Computer Aided Design (CAD) model to improve quality. Engineers can develop an inspection sequence to check critical dimensions from the CAD model governed by the engineering drawing or specification. Machine vision then assesses the part against the model in-line and either approves the component or rejects it to the scrap bin. The added data from the third dimension moves the inspection process from a 2D check against the drawing to a direct comparison between CAD and product.
3D-machine vision delivers improvement to manufacturing quality, throughput, and cost. Ideal for high-volume production steps, 3D-machine learning adds depth to the previous 2D approach to create a complete picture of the part. This image can determine the size, shape, location, and position of a defect, provide that to inform an algorithm to improve processing efficiency or improve supply-chain inventory turns through scanning, picking, and reordering. Machinery using this disruptive technology can absorb unexpected variables and obstacles, navigate them and complete their tasks without reprogramming.
Adam Kimmel has nearly 20 years as a practicing engineer, R&D manager, and engineering content writer. He creates white papers, website copy, case studies, and blog posts in vertical markets including automotive, industrial/manufacturing, technology, and electronics. Adam has degrees in chemical and mechanical engineering and is the founder and principal at ASK Consulting Solutions, LLC, an engineering and technology content writing firm.