Update your Existing Quality Assurance Process from Manual to Automation

By updating a Quality Assurance (QA) process from manual to automation using AI/ML algorithms, your process will become more efficient, increase accuracy, scalability, and overall improve process performance to keep you ahead of the competition.

At ADS, we have been helping our customers extend the life of their systems, processes and products by providing cost effective automated solutions for over 25 years.

In this month’s Data Mail we discuss:

• Why go from manual to automated
• Add Intelligence to Your Quality Assurance Process
• Vision System for Defect Detection Examples
• Talk to us! Find out where we will be over the coming months

Contact us today to discuss how ADS can help you extend the life of your existing investments by automating your systems instead.

-Steve Seiden – President, Acquired Data Solutions

WHY GO FROM MANUAL TO AUTOMATED

For nearly all of human history, manufacturers had to rely on manual inspection to perform visual quality inspections. However, with the advent of computer vision and the recent addition of machine learning, manual Q/A tests have become easier than ever to automate. Here are some of the key disadvantages of manual Q/A:

1. Accuracy: Humans may miss important features, risking false results.
2. Repeatability: Humans have different levels of attention to detail, creating the risk of conflicting Q/A results between different operators
3. Reliability: Environmental and emotional factors can impact a manual operator’s performance from day-to-day
4. Throughput: The only way to improve throughput is to add more operators
5. Scalability: Adding new operators requires investment in training and employee benefits

Augmenting or replacing manual Q/A processes with CV systems addresses all of these concerns and lays the groundwork for future improvements. CV systems are highly scalable, support much higher throughput rates, and inherently lack the limits to accuracy, repeatability, and reliability that humans possess.

Add Intelligence to Your Quality Assurance Process

Transform-CV is ADS’ Deep Learning-enabled computer vision application for Quality Assurance applications. Transform-CV analyzes images and makes real-time inferences, providing customers with crucial insights into product quality to improve manufacturing processes. Based on ADS’ value-adds to open-source software, Transform-CV is readily tailorable to a range of customer applications. Transform-CV is based on commercial hardware and supports both PCle and PXle platforms to address a wide range of manufacturing test & measurement environments that may be employed to augment or replace manual defect analysis, dimensional analysis, and component classification. Below are three application examples.

Find out More

Vision System for Defect Detection

Below are a few examples of how we used our Transform-CV for Quality Assurance Applications.

?Microelectronics

Transform-CV for microelectronics is a vision system designed for a manufacturer of high-power electronics to evaluate wire bonds on circuit boards. For every bond present in-frame, Transform-CV makes real-time inferences to determine the bond’s class and whether it’s attached to a connection wire. The system also geolocates each bond and creates a global map, which is referenced against a map of desired locations within a specified tolerance. The system displays a count of the bonds and their attributes in real-time. Due to the extremely small size and reflective surfaces of the boards tested, Transform-CV operates at 50x magnification and features polarization.

Outputs:

• Bond type (spot, ball, wedge, ribbon)

• Bond connection status (connected/unconnected)
• Bond geolocation (within tolerance)

Precision Machining

Designed for a precision machine shop, Transform-CV evaluates machined plates for the presence of scratches, dents, and abrasions. This system features a motorized XY stage to move plates along a predetermined path under a high resolution camera, and notifies the operator of any surface defects larger than the rejection criteria, while the operator still makes the final PASS/FAIL verdict. Due to the small size of the defects (rejection criteria of >0.01 in) and large part footprints (plates are 1ft x 1ft), the system operates at 10x magnification and utilizes software-stitching to compile multiple images into a single image of the entire plate.

Outputs:

• Presence of defects
• Location of defects

Component Classification

Developed to demonstrate the potential of machine learning-enabled computer vision at trade shows and conferences, Transform-CV.1 performs component classification in real-time while occupying a desktop-sized footprint. The system features an industrial camera, illuminator, rotating turntable, GPU-embedded computing platform, and a mixture of screws, nuts, and washers. The camera records the hardware scattered atop the turntable, whose rotation simulates a moving assembly line, and the system identifies and counts the number of each hardware type present. The system displays the hardware counts, whether the counts match the user-defined quantities, confidence level, parts per second supported, and framerate.

Talk to us! Find out where we will be over the coming months

We would love to discuss your projects with you in person. Please find a list of trade shows that we will be attending over the next few months. Alternatively, contact us if you just can’t wait and would like to arrange a site visit:

• ITEA Multi-Domain Operations Workshop | Alexandria, VA | July 16-18, 2024
• IEEE Autotestcon | National Harbor, MD | August 26 - 29, 2024
• International Manufacturing Technology Show | Chicago, IL | September 9-14, 2024
• Air Space and Cyber Conference | National Harbor, MD | September 16 - 18, 2024
• ITEA Cybersecurity Workshop | Orlando, FL | September 17 - 19, 2024