A vision for zero defects
Canadian PlasticsAutomotive Canadian Plastics Software
Want to analyze your root production problems and eliminate bad parts entirely? Fredericton, N.B.-based industrial consultant Eigen Innovations has a machine vision software platform that can help.
When it comes to part quality testing in manufacturing, the jury is still out on whether destructive or non-destructive testing is the better method. But what if a production process could adjust its process controls in real-time, and continually improve, to prevent any quality issues in the first place? The quest for zero defects is the ultimate goal of any part maker, and a Fredericton, N.B.-based industrial software platform developer wants to help them move a step closer.
Eigen Innovations Inc. was founded in 2012 by a team of machine learning, mechanical, and software engineers trained at the University of New Brunswick (UNB) in Fredericton and Saint John, and the 20-employee-strong firm is now making its mark as a supplier of a platform that unites machine vision with process and quality data, primarily for automotive part suppliers. According to James Finch, Eigen’s director of customer success, the firm’s artificial intelligence (AI) machine vision platform has been purpose-built and optimized to analyze factory image and process data as it’s captured, creating a traceable digital profile for every factory part and component in order to analyze root problems and eliminate defects altogether.
“Our hybrid cloud and edge software suite is designed specifically to address complex challenges facing high-volume part manufacturers, in particular challenges relating to plastic welding and complex surface vision,” Finch said. “It features proprietary image processing and data extraction techniques that transform vision systems from stand-alone inspection tools to an enterprise-level, process intelligence platform that can make changes on the fly, shut the machine off in less than a second if it detects bad parts being made, and provides insights on how to prevent that defect in the future. The goal is to give our customers the ability to see inside their products and processes and to use that knowledge gained to overcome some of their biggest production challenges.”
MAKING A DIGITAL RECORD
Typically, an Eigen factory configuration includes camera sensors to capture a critical view of parts in process; an Eigen smart module that collects data from camera sensors and PLCs and deploys machine learning models; an HMI monitor and label on the factory floor; an online cloud platform to monitor, analyze, and collaborate in-factory or remotely; and a PLC integration to automate actions when process or quality issues are detected. “Our platform gives the customer’s operators and engineers real-time information so they can quickly adjust process controls to prevent quality issues inline,” Finch said. “They also have access to advanced data analysis tools, so they can quickly access, search, and compare parts to conduct root-cause analysis and investigations.” When deployed on multiple machines in multiple factories, Finch continued, the platform drives enterprise-level process control and improvements that reduce costs associated with poor quality. “The digital record we can produce lets us create a consistent blueprint to measure from, even if the camera angles aren’t perfect, and we’re able to take raw digital images and overlap them on the digital blueprint of a part to make sure that part is meeting its quality standards,” he said.
In one recent case study, Finch said, a plastic fuel tank manufacturer had been using traditional quality monitoring and destruction testing to satisfy quality standards on a critical welding process, but wanted an upgrade. “We installed a thermal camera machine learning solution that monitors critical parts along the weld leg and prevents parts with inadequate welds from moving beyond the weld cell, massively reducing the risk of defective parts leaving the factory,” Finch said.
And among its other customers, Finch said, Eigen has decreased the reject rates at one shop from eight per cent to less than one per cent, and saved another hundreds of thousands of dollars by eliminating destructive testing on a critical plastic part.
THE ROAD AHEAD
Despite its successes, Eigen hasn’t forgotten its roots: The firm continues to recruit new staff from UNB and maintains a manufacturing line in a laboratory on campus to test its solutions.
And Eigen also knows where it wants to go in the future. “For 2022, we want to unlock scalability by partnering with a machine builder that can sell our solutions built into its own machines, and we’re also looking for deeper partnerships with our own customers,” Finch said. “And while most of our solutions currently involve high-volume welding applications for large enterprise clients in the auto sector in Canada and the U.S., in the longer term we’re looking at expanding our partnerships into South America and Europe and integrating our solutions onto any high-volume manufacturing process in any sector.”
In a market that’s now flooded with defect detection providers, Finch says there are good reasons why Eigen stands out. “Our differentiators are that we can not only detect part defects, we can prevent them from happening in the first place with the process adjustments that we provide,” he said. “We want to get our customers to the place they all want to be: zero-defect manufacturing.”