The Impact of AI and ML Inspection Systems in Manufacturing

Advanced technologies are continually optimising manufacturing processes, increasing the speed of production and improving quality. One area that is particularly primed for improvement is in the inspection of products as they move along the line. Vision detection and inspection systems using AI and ML can further streamline operations, minimising human error and increasing accuracy, as electronics manufacturer Flex found when it implemented AI/ML inspection processes at multiple sites in 2022.

“The technology helped greatly reduce inspection time, improve quality and, as a result, increase production efficiency,” said Rahul Katkar. “These efforts also supported other initiatives, such as closed loop controls and inspector upskilling.”

Here, Katkar tells us more.

AI/ML inspection systems enhance efficiency and empower workforce

The systems also delivered significant cost savings, reducing scrap by identifying issues before sending a part to another step in the line. At Plex, the tools helped diversify the skill sets of team members - who, rather than losing a job to AI-based automation, were trained to use the tools to create new AI/ML models, thus furthering their careers. 

“The lessons learned from Flex’s implementations, which have led to further use of the technologies, can help inform companies looking to expand their use of AI/ML on the shop floor,” says Katkar. 

The traditional manufacturing inspection process involves human workers examining products as they move down the production line. But automation, robotics and other advancements that come with Industry 4.0 technologies have increased production speeds to the point that quality inspections are difficult for humans alone to perform consistently. 

“In addition to dealing with the speed, inspection criteria can cover a wide range of items, from screws and wires to labels and other vital components,” says Katkar. “After performing inspections for many hours, visual fatigue can lead to human errors.”

To combat this challenge, Flex developed two options for AI/ML-based detection and inspection systems purpose-built to improve quality checks on the factory floor. The system’s trained neural networks can detect defects that are difficult for human inspectors or conventional vision systems to find. 

“The AI/ML systems also learn on the job, steadily improving performance,” Katkar continues. “Even if a system doesn’t appear to improve efficiency immediately, its capacity to learn will pay off over time.”

To start the process, engineers need to label and train the models based on the products and systems they are analysing. From here, the engineers should display and generate the inspection based on photos sent to the AI model. Finally, before deployment, teams must evaluate the results to ensure there are no false calls or errors caused by poor data or training. 

“Once confidence levels are high and the capabilities are programmed, the solution can be leveraged for key error groups like image classification, anomaly detection, object detection and segmentation.”

Katkar says that once rolled out on the shop floor, the results are promising. In fact, at one Flex site inspecting the production of hardware, the customer saw a 30% improvement in efficiency and 97% payback on its initial investment in less than a month. At another location, involving the production of products with sheet metal components, the customer saw an efficiency gain of 28% and a triple digit return on investment.

“In addition to increased production and cost savings, an AI/ML inspection system delivers other benefits, perhaps the most important being the impact it has on the employees,” he says.

Rather than replacing workers, the AI/ML system creates opportunities for the inspection staff to up-level their skills to manage the new technology instead of continuing to perform onerous inspections. 

“Workers are also freed up to focus on more strategic manufacturing operations. This can boost morale while enhancing employees’ career paths,” he adds.

Revolutionising manufacturing with AI and ML

A broad debate is ongoing about the implications and risks of AI to society, but in manufacturing, AI and ML are finding their place in practical optimisations. Companies that don’t take advantage of these technologies risk being left behind. What Flex has found with its implementations is that AI/ML systems not only increase efficiency and reduce costs, but they create new career opportunities for workers. 

“AI and ML are at the forefront of Industry 4.0 efforts to transform operations,” says Katkar. “Using AI/ML inspection tools is a very effective way to improve results and increase production across an entire enterprise.”
 

****** 

Make sure you check out the latest edition of Manufacturing Digital and also sign up to our global conference series - Procurement & Supply Chain 2024 & Sustainability LIVE 2024
******
Manufacturing Digital is a BizClik brand.