Zero-Defect Manufacturing: Are We There Yet?

From cosmetic flaws to machine malfunctions to production line errors, defects that occur on the factory floor are an ongoing and costly challenge for manufacturers. Smart industrial computer vision solutions are now coming to the rescue, as manufacturers tap AI-enabled machine vision that can help them move closer to zero-defect manufacturing.  


Article Key

Image credit: Eigen Innovations

Every manufacturer wants to obtain the highest quality throughout the production process. However, failures can occur in multiple ways, including human error, failure of the process itself, or malfunction of the equipment used on the factory floor. 

Fortunately, the rise of Industry 4.0 has ushered in more intelligence and IoT automation in manufacturing, including smarter computer vision solutions that can help companies investigate, capture, monitor, and analyze industrial processes. One such company moving automated defect detection forward is Eigen Innovations. Based in Fredericton, New Brunswick, Canada, it specializes in helping manufacturers realize zero defects by giving them real-time information about the quality of their products and their manufacturing process.

Eigen’s industrial software platform is designed to enhance image data by correlating it with machine/process data. By doing this, manufacturers have the ability to detect quality issues in real-time, which leave them better equipped to prevent manufacturing issues and equipment malfunctions from occurring in the first place.

Combining Virtual Images with Process Data

Eigen Innovations platform was developed to help manufacturers extract insights from raw vision data. The solution takes the image data and creates a virtual image of each part. The virtual image–combined with the real-time process data–gives manufacturers a traceable profile of each part. The software uses the profiles to build machine learning models for defect prevention and optimization, allowing quality engineers to quickly review parts in-process and adjust process controls to prevent quality issues. 

Eigen Innovations has partnered with Intel® to optimize the performance of its machine learning models. While many vision solutions provide part inspection or process monitoring, Eigen builds on the Intel® Distribution of OpenVINO™ toolkit to offer manufacturers an inline view of quality and process indicators, delivering real-time insights, monitoring, and remote configuration services. 

Using proprietary image enhancement techniques and image normalization processes, the Eigen platform reduces the need for highly customized inspection algorithms. Listen to our IoT podcast episode with Erin Barrett, Chief Revenue Officer at Eigen Innovations, to hear more about how this technology came about and how it works.

Shining a Light on QC in Automotive Manufacturing

The Eigen Innovations platform is geared toward various industrial sectors where high-volume, high-value parts are produced or where conventional vision solutions are unable to help prevent complex quality issues. 

One such sector is in automotive manufacturing, specifically in the area of car lighting systems. Recent innovations in headlamp, signaling, and light sources combined with evolving styling complexity, have added massive value to automotive lighting assemblies. Ensuring quality, performance, and safety while obtaining access to global markets is challenging. Companies that manufacture head and tail lamps need a high level of certainty in their welding processes to ensure they can meet stringent OEM quality standards.

Eigen recently helped a large Tier 1 automotive manufacturer detect and prevent issues along the weld leg that were resulting in costly defects, such as leakers. Prior to moving to the Eigen integration, the manufacturer experimented with various rules-based vision solutions that were inadequate in detecting issues along the weld leg.

According to Eigen, the manufacturer’s reliance on destructive testing meant data was only being collected on those destructed parts. That process left the manufacturer uncertain about the weld quality across all its parts. Due to the complexity of exterior lighting parts, a poor quality weld could lead to the entire assembly needing to be replaced, which was costly.

black and purple automobile lights

Image credit: Eigen Innovations/automotive lighting imaging system

Image and Process Monitoring

The manufacturer worked with a local solution integrator to incorporate Eigen image and process monitoring. The manufacturer set up multiple thermal cameras to capture views of the welding process, which are used to generate virtual part images for part-to-part monitoring.

Edge computing devices ingest the image and process data, which is correlated to each virtual part profile. Machine learning models monitor critical parameters along the weld leg, and the platform can alert operators when issues are detected.

According to Eigen, these insights delivered through its software platform prevent parts with inadequate welds from moving beyond the weld cell. This process massively reduces the risk of defective parts ever leaving the factory.


Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.