Automation Fair Q&A: Inside Rockwell’s Push Toward Autonomous Packaging for CPGs

It’s highly orchestrated. We still absolutely think people will be in the mix. We’ll move people from mundane tasks to orchestrators overseeing processes. Autonomy is really about using all available information in real time, either open loop or closed loop. Customers can be anywhere on that continuum — diagnostic, predictive, prescriptive, or autonomous.

Autonomy is using all the data signals to determine the optimal next step and then continuously executing on that step. You can put a human in the loop or automate back to the PLC through closed-loop control.

Let’s talk technology. What specific new products or tools are helping brand owners move toward autonomy?

In our booth we show all the different applications in the plant where we’re using AI to drive autonomy. On the engineering side, we’ve got FactoryTalk Design Studio doing codedevelopment. For packaging, Vision AI is also very relevant as it leverages AI (instead of rules-based algorithums) to spot defects in-real-time.

Is Vision AI the same thing as Optix or something different?

No, it’s different, it’s called VisionAI. You run it via an industrial PC at the edge. It’s a cloud-based system. You can aggregate images to the cloud — low-code, no-code, easy for OT people. You label what’s a defect and what’s a good product, let the AI engine learn over time, and then deploy back to the edge.

It’s integrated with the Logix control system, so you can get to closed-loop control. Once it sees an issue, it can act on it in real time.

You presented on Perfect Fill earlier today, and it was a big splash presentation last year.  Can you walk through that, and where predictive and model-predictive tools fit in?

Some customers are just looking to identify where their yield loss is occuring – which is where an MES system can help. MES is diagnostic. The next level of maturity would be predictive solutions - something like LogixAI, which predicts fill weights and adjusts in advance to get closer to perfect fill.

Then the third level is autonomous control solutions like Model Predictive Control (MPC). It optimizes across multiple systems. We’ve used it with frozen foods where inputs vary — size, moisture content. You’re balancing moisture, throughput, quality. MPC is the algorithm that handles multiple unit operations. If helpful, here is an analogy: MPC is like smart driving. Most peoople want to get there quickly, safely, minimize fuel usage, and maintain a smooth ride. Cruise control helps but doesn’t anticipate hills – so it overcompensates, hits the gas – wasting fuel and jolting the car. MPC looks ahead, sees the hill, adjusts early to optimize performance across multiple variables (in this example speed, quality, safety, energy). It used to be called Pavilion8; it’s now PavilionX.How Model Predictive Control reduces variability: Pavilion MPC smooths process variability by moving operations closer to specification limits while maintaining safety margins. Benefits include reduced variability, “plant obedience,” constraint management, and overall uplift in performance compared to traditional operator-controlled processes.Rockwell Automation

We’re also hearing about copilots and small language models. What’s happening there?

We’re embedding generative AI through copilots within Optix. We partnered with Microsoft on small language models and integrated copilots within the HMI. It’s a frontline interface for operators to interrogate and ask questions before they jump to a maintenance order.

You can feed it manuals, OEM documentation, 10 years of maintenance notes — whatever you choose. Some customers add their weekly reports, and the system begins seeing correlations. The more data you feed it, the better the results.

Any other autonomous-leaning tools CPGs should know about?

Guardian AI is another. It autonomously predicts when you’ll have a failure in a motor, fan, blower, etc. It uses PowerFlex Drives as sensors, monitoring electrical signals to identify anomalies and flag maintenance. Over time, you train the model on what’s a true failure versus what’s not.

Shifting gears — what are CPG brand owners asking you about this week at the Fair? What’s top-of-mind for them?

In our booth we show timely customer challenges across design, operate, maintain.

In design, there’s pressure around reformulation — for instance, the “Make America Healthy Again” (MAHA) movement. We highlight Composable PLM from Kalypso. Most CPGs already have PLM systems, since CPG is a fairly mature industry. But they’re expensive to rip and replace. So we offer a composable architecture — an intelligence layer using chatbots or generative AI, acting as agents across systems.

Digital twins help with the engineering-to-operations handoff. If a functional beverage can’t run on an existing soda line and needs a tunnel pasteurizer, the twin shows how that affects the whole line.

There’s pressure to innovate because of private label. Companies still have more SKUs but are rationalizing, focusing on hero SKUs.

Many brands, say Mars' KIND bar brand,  are shifting from polypropylene flow wraps to paper-based materials. Would Composable PLM help manage that transition?

Exactly. Packaging management would be one of your spec-management jobs within PLM.

What about the operate/maintain side?

Workforce and cost control are huge. We’re showing AI solutions that augment the workforce to reduce costs and improve margins. Customers are super price-sensitive. Margin expansion has to come from the bottom, from cost reduction, not from the top. We see consumers trading down.

We’re also seeing a lot of investment in modernization. There’s uncertainty in regulatory policy, so many customers are modernizing their base. Services, distributors, partners — we try to make modernization easy so they can layer on the fun stuff afterward.

Many engineers know they need to modernize but struggle to justify it internally. You’re helping with that, right?

Yeah, we built an ebook: “Building the Business Case for Modernization.” Everyone knows they should modernize; the question is why it doesn’t happen. It’s hard to get justification when you’re competing for limited dollars. We highlight how to build a case that goes beyond cost and risk — focusing on value accelerators.