April 12, 2024
Computer Vision and AI

Manual processes in the era of AI: How computer vision gives manufacturers an edge

Manual processes are still a core part of the manufacturing industry. But advances in AI and computer vision hold the key to operational efficiency and digital transformation.

The state of manual processes in the manufacturing industry

Despite what many would assume, manual processes are still a core part of the manufacturing industry. From manual assembly and machine setups to material delivery and logistics, human work on the shop floor is critical to delivery, and for many industries, that’s unlikely to change soon.

So there’s huge potential in identifying where these manual processes can be optimized and improved upon. But when it actually comes to figuring out what improvements can be made on the floor (and where), things become more difficult.

Data collection on production lines

To start, there’s not a lot of tracked data for manual processes. While shop-floor data collection systems such as manufacturing execution systems (MES) can help monitor and control shop floor production, they don’t explain how processes are happening, or why issues are occurring. Companies seeking this information then need to deploy specialized engineers to painstakingly record data in real time.

Unfortunately, collecting data on the floor takes a long time, and is often unreliable. Even if you can dedicate headcount to just focus on observation and data gathering, information can be incomplete or error-prone. Observations also rarely document what happens day to day, and even worse, decisions made based on faulty data are also unlikely to address underlying problems.

Assessing the effectiveness of improvement measures on the production line can be difficult. Incomplete data and poor decision making create an endless feedback loop of slow deliveries and inefficient processes, and it’s hard to unspool the tangled threads of problems to figure out where issues originate.

You're all my Lucy.” – Will & Grace Recap – We Love Lucy

Collecting data is a pain: It’s time consuming, and it’s hard to capture the big picture. Traditional methods can’t always keep up: manual observation takes ages (and wages). Data tracking is often poor, and additional specialized personnel are needed for all steps of the process. And if you’re spending all of your time on granular data collection, it’s hard to find any time to spare on actually streamlining your operations. 

It’s also hard to identify focal points for improvement. Where do you send your continuous improvement managers? Do they monitor individual workers? Do they count mistakes on the line overall? Do they jump from station to station, possibly missing things in the process? Unless you can deploy a committed team (expensive) or a staff member with a thousand eyes, (dear reader, given today’s shortages in skilled labor, such mythical creatures are in short supply), you’re out of luck.

Index - Kultúr - Egyszer sem látta a Gyalog galoppot? Elmeséljük  animgifekben!
Pros: He’ll have a great view of the shop floor. Cons: He’s got a hankering for human flesh that the factory canteen can NOT satisfy.

Let’s face it, even the best of teams can use a little bit of extra help.

Modern problems demand modern solutions

So, if massive teams of line monitors aren’t feasible, and many-eyed monsters bring bad vibes to the shop floor, production managers and process engineers need a better way to gather actionable insights. That’s where computer vision and AI technologies come in.

First, it’s important to understand what we really mean when we use the term “computer vision.” IBM defines it as “a field of artificial intelligence that uses machine learning and neural networks to teach computers and systems to derive meaningful information from digital images, videos and other visual inputs — and to make recommendations or take actions when they see defects or issues.”

This has huge implications for the manufacturing industry. Cameras on the shop floor can monitor many stations at once, and collect and record thousands of data points. In a recent industry report from logistics giant DHL, the company acknowledges the growing importance of exploring the applications of computer vision in logistics, manufacturing, and beyond.

 “With advancements in artificial intelligence (AI), computer vision is at a stage to become an industry-shaping technology and has exceptional promise along the supply chain – for our customers, employees, partners, and certainly the environment.” – Katja Busch, Chief Commercial Officer DHL & Head of DHL Customer Solutions and Innovation

In the past, computer vision applications were mostly limited to the area of quality control. Cameras on the floor could assist with defect detection, assembly verification, label verification, and so on. However, tools that effectively combine computer vision with the processing power of AI have the potential to improve operations in a much vaster variety of areas, including:

Streamlining processes

Computer vision-based platforms can help document and digitize manual work, providing analytics on efficiency and suggesting ideas on reducing downtime to get more out of current work.

These systems can also catch problems as they happen, sending real-time notifications to plant managers, who can adjust or stop work before too much time is wasted. 

Enhancing worker capabilities

There’s a shortage of skilled labor in many industries. Implementing a computer vision solution can help supplement expertise by providing training and assistance.

Computer vision systems can also provide real-time feedback to workers, highlighting areas for improvement. This feedback can prompt workers to adjust their techniques or behaviors to achieve better results. Instant alerts can also ensure that production lines meet safety standards, keeping the people on the floor safe from accidents and other errors.

Over time, an AI-based solution can also create a library of best practices videos, to be used for worker training and education. 

Improving overall operational efficiency

Computer vision platforms learn over time, suggesting better, more specific improvements. The longer cameras gather and record data on the shop floor, the more specific the suggestions can be. Longer-term implementation means better insight into systemic inefficiencies. 

Computer vision systems also catch problems as they happen, sending real-time notifications to plant managers, who can adjust or stop work before too much time is wasted. This makes it easier to hit production targets time and time again.

Computer vision and technical debt

Not all solutions are created equal. Companies looking to digitally transform their shop floor operations often don’t have the time or money to take on a lot of technical debt in the form of complex solutions. 

The common assumption is that integrating new technologies into existing manual processes is complex and expensive. Often, solutions require significant changes to infrastructure, workflows, and employee training.

The benefit of computer vision is that it’s relatively simple to install (even on a single production line), and requires minimum hardware to get going. 

The ideal product should also be easy to use. AI can do the computational and analytical heavy lifting, providing users with easy to understand graphs and charts, and suggests clear, actionable improvements. The learning curve is minimal, allowing for easy onboarding for process engineers and other employees in charge of monitoring output and efficiency. Time that used to be spent on in-person monitoring and data gathering can instead be spent on improving processes, planning better employee training, and optimization. 

Companies can also easily test a computer vision solution on a single production line, and then expand it to other lines as they get comfortable. A good computer vision solution can gather the data it needs within days. For production lines with shorter cycles, the AI can start identifying patterns within one to two shifts. No need for supervision or labeling either. That means proof of concept happens quickly, without the need for a massive upfront investment. 

Finding the right solution for your shop floor

Manufacturing companies that rely heavily on manual processes don’t necessarily have to spend a lot of time and money on digital transformation. Operational efficiency is within reach.

Deltia offers an AI-based process analytics and monitoring platform, along with camera installation on your shop floor. No fancy cameras, complex calibration, or super special lighting conditions needed either. The Deltia product provides insights into how to improve manual operations, and is built by people with a deep understanding of the day-to-day problems of the factory floor. The result? Real-time transparency and deep insights into how to improve efficiency and quality in manual processes.

Let us tell you more about how Deltia works in person. Click here to book a demo with us today, and learn more about how Deltia can help your company meet its productivity goals.

Want to learn more about Deltia?

Sign up for a quick demo.

Book A Demo