The One Question You Should Be Asking Before Deploying a Machine Vision System (Especially If You’re an Automaker or Electric Battery Supplier)

Just curious: How many questions did you ask about software the last time you bought a machine vision system or considered using machine vision to automate quality inspection?

Probably just these two, right?

1. How much will the software license cost me to run the hardware?

2. What’s it going take to integrate, manage, and train users on the new component? Which I translate to mean: “How difficult will it be to deal with the software that you’re telling me I need to run the hardware?”

While these are some of the right questions to ask, I bet if you had asked just one more question, your job would be a lot easier than it is right now.

3. Is there a way to run my entire system from the same vision software suite (i.e., all the software is provided by a single vendor) even if I need vision hardware from different vendors?

This third question is the one you should ask first. It’s the question that will help you decide whether the money you’re about to spend will finally fix – or help you avoid – the most frustrating thing about traditional machine vision system designs: the hodgepodge of software platforms from different vendors and the challenges that setup poses when you want to add or update hardware components.

I say that because software is the glue that underpins these systems. It either enables or prevents the different hardware components from coming together in the way you need to automate quality inspections.

Whether you’re trying to improve quality control or conduct compliance audits, software is what…

• connects your visual inspection technology with other business systems to support automated data capture and analysis.  

• your operations and IT teams will have to interact with all day, every day, to engineer inspection processes, possibly train deep learning models, decide what passes or fails inspection, meet service level agreements (SLAs), and protect your brand’s reputation for quality and safety.

• has given machine vision – and automation, in general – the reputation of being “complicated, expensive, and frustrating” over the years.

Many people assume that using hardware from different vendors is the major challenge when building a vision system, but hardware isn’t the problem. What’s tricky is getting the mix of software packages from different vendors to work well and with one another that gets tricky!

So, make sure you scrutinize the software underpinning your machine vision system, especially if you’re an automaker or supplier who is gearing up for increased demand for electric vehicles (EVs).

Don’t let someone tell you that you must use software from five or six different hardware vendors to be able to automate quality inspections. You don’t.

There are user-friendly, scalable, and secure single-vendor software suites that can …

• run different types of hardware components such as cameras, barcode scanners, sensors, and robotic arms for visual inspection, process automation, and track-and-trace.

• run cameras, sensors, and scanners, from different vendors into the system without operators needing to configure, learn or manage different software packages or pay licensing fees to multiple vendors.

• interface with the systems that run robots to help guide them in automated workflows.

• run your entire machine vision system, end to end, however you need to use it.

Avoid Regrets and Protect Your Reputation

I realize that you – like many automotive original equipment manufacturers (OEMs) and suppliers – likely turned to machine vision in the past because you wanted to make it easier to meet stringent production tolerances, ensure precise quality control, and keep pace with evolving industry standards. So, you probably pieced together a system over time as you needed new capabilities to satisfy increasingly complex production demands, maintain consistency with your inspections, and reduce defects. It’s difficult to maintain higher throughput and regulatory compliance without automating. But in the effort to gain efficiency and ensure repeatable accuracy, you ended up with a collection of various software packages, along with a never-ending headache.

That’s okay. The growing demand for EVs is giving you a very rare opportunity to reset things – or ensure you get things right the first time if you’re new to the industry.

Machine vision doesn’t just have to be a way to compensate for labor shortages or validate your workers’ decisions. Machine vision can be the mechanism you use to build trust in your brand and protect your bottom line. It can help you be confident that defective parts or vehicles aren’t slipping past inspectors and help you intervene quickly when issues are identified. It can be an accountability tool.

But you must be able to make incremental changes to your machine vision system as business needs change.

So, next time you have a conversation about automating inspections, ask the tech provider or system integrator:

“Can the system design you’re recommending be set up to run from a single software suite, even if I use different hardware vendors?”

If they say, “Yes,” then you can start asking about the software’s interoperability/compatibility with everything else in your tech stack, data security options, and how it impacts the overall scalability of your operations:

• Does this software meet our IT-defined security requirements? And how easy will it be to verify that?

• Is the software backward and forwards compatible with all vendors’ hardware?

• If I expect that I’ll need to add on more cameras in the next few months, what will that entail?

• Will the software support 2D/3D inspection, deep learning-based OCR, or other emerging technologies?

• Does the software allow me to train AI models to help with image analysis and automated decision-making? Or will it incorporate AI in the future?

• How often does the vendor generally update the software? Will there be support offered as the tools continue to be developed?

• How long will it take to train people on the software? Will they need retraining every time we change the process or hardware? Do they need any specialized IT or data science knowledge?

• Will one software license support all use cases and extend across all my facilities?

• What other types of technology can I program and manage with this software, besides traditional machine vision cameras/controllers?

• I know the entire software suite is from one vendor, but will users need to pull up different dashboards for each software platform or application?

Don’t Accept “No” as an Answer

Now, I realize some people may tell you that you can’t run your entire machine vision system using a single software suite from one vendor, a la Microsoft Office. They may tell you that you must use the software provided by each hardware component’s manufacturer or say that your manufacturing execution software (MES) is actually “the glue” for your inspection process.

But my team and I can show you how to build a machine vision system in which you only have…

• One company to deal with for the machine vision system design, integration, maintenance, and innovation. It’s important you work with software vendors who are committed to working with you long past the initial go-live and who will keep the software updated to accommodate your evolving needs.

• One software suite for your team to learn. You wouldn’t ask them to use Google Docs for some tasks and Microsoft Word for others. So, why use software from different machine vision vendors when you don’t have to?

• One software suite to manage for the life of your machine vision system, which could be the life of the production line – and future lines. If you need 2D/3D cameras, want to incorporate AI, or do something else more advanced in the future to automate your workflows and build trust in the automated decisions, you will have the flexibility to do so if you choose the right software  today.

We’ve done this for several automakers and suppliers over the years, and we’re doing it now for EV OEMs, including electric battery suppliers. We’ve also done this for others outside the auto industry, such as KINE (food and beverage production) and STIHL (power tool manufacturer).  

So, if you’re feeling stuck or have questions, don’t hesitate to reach out. We’re here to help you find the right “glue” for your system based on your objectives, budget, and operational design.

Remember…

Building a machine vision system is like constructing a house. If you invest in the right software foundation now, it will save you from constant remodeling later.

Swapping out hardware is easy, and the right software will make future workflow updates feel like minor tweaks rather than major overhauls. So, give yourself room to grow. Focus on finding software that can run any hardware you want and won’t leave you with regrets in a couple months or years. And find someone you can trust to ask and answer the right questions, starting with software-related questions.

###

On a Related Note

You may also want to hear what’s shared by other vision specialists here…

• Industrial Automation Insider: What You Need to Think About – and Be Prepared to Change – When Spec’ing Out a Machine Vision System
• Everyone Says AI is Going to Make Machine Vision – and Electric Vehicle Quality Inspections – ‘So Much Better’ – and It Will. But There are Other Things that Will Make a Bigger Impact.

As well as these stories from people who have found effective ways to use machine vision technology to solve lingering pain points and save money in the process…

• MOSAIC Refines Quality Assessment with Vision System
• KINE Robotics Reduces Costs with Accurate Conveyor Tracking System
• Machine Vision Technology Helps STIHL Cut Cost, Time and Human Error