By Liz Stevens, writer, Plastics Decorating

As manufacturing companies evaluate their use of automation, questions arise on how well these innovations work in plastics decorating and how they are most effectively deployed.

At the 2025 Plastic Product Decorating Summit, the panel discussion “Efficient Operations with Automation” tackled these questions. The session featured insights from Michael Sansoucy (Arburg, Rocky Hill, Connecticut) and Mark Patterson (FANUC America Corporation, Rochester Hills, Michigan). Sansoucy and Patterson discussed the technologies offered by their companies, as well as other technologies making waves.
 
Where are companies seeing the biggest impact with the addition of automation and robotics?
Sansoucy: The biggest impact of automation is in the area of workforce, because of the lack of qualified labor in the industry. When it comes to printing or direct-to-object or labeling, it is impossible to do this work, accurately and repeatedly, with a human hand. So, automation is a critical need for in-mold labeling or thin-wall packaging. There is no human who could keep up with the speed of those operations, so automation is mandatory. Another major driver is increasing operational efficiency, and we are addressing that with automation.
 
Patterson: I agree. The primary driver for automation in plastics remains labor availability, followed by quality and throughput requirements. People can’t find workers. That is a problem everyone has right now. And then, when they do find people, the new hires need extensive training to run equipment, place components where they need to be and avoid high scrap rates.

How are companies assessing the ROI of automation?
Patterson: When plant owners consider automation, it can help them with the labor issue and with product quality. Typically, customers evaluate automation based on total labor cost. However, ROI discussions increasingly include throughput gains, scrap reduction and quality improvements. These perspectives allow customers to broaden the scope of their ROI and even the time that they’re willing to wait to achieve a return on investment. The acceptable ROI window has expanded from one to two years, up to two or three years in some cases.
 
Sansoucy: The big challenge is in addressing the high-mix, low-volume operations. Automation has set itself up well for high-volume, low-mix but not for the opposite scenario. In-mold labeling is a great example of the challenge in packaging. Twenty years ago, a customer bought a machine, one mold and a robot, and the plant floor set-up ran that one packaging product. The customer never had to touch that production line, and it was easy to justify that purchase. But today, the typical situation requires running an injection molding machine on three or four different containers with another 10 SKUs for labels, plus maybe lids and maybe wrap around. Those customers are coming to us and asking how to design a system that can run 20 products. That is where the return on investment can be an even greater factor in deciding whether to add automation.
 
What questions do you ask to begin the discussion on automation and robotics?
Sansoucy: We have a two-page checklist that we go through, but mainly we want to know what will happen with the product at the end of the run. Is it going into a box or into a tray, and how will it be handled? We also want to know where the mold is coming from. Does the mold exist? What kind of space constraints are on the plant floor?
 
Another big item on the checklist is asking about the expected investment from the customer. It is possible to automate anything, given enough time and money. But if the customer has a $200,000 budget and a three-month timeframe, and we present them with a $2 million solution that will take a year to implement, that’s not going to work.
 
Patterson: We use two types of checklists to ask questions. The first checklist helps to determine the viability of automating a job in the plant’s overall operations. We need to understand the larger process, the job’s run time and the volume of parts to be run before a changeover occurs on the machine. For example, I often work with customers who say they run 1,000 parts a month. My first question is, do they run 1,000 all at once? If they are running 100 this week, and then 20 next week, this is not a job that can be automated. But if the production can be batched together to run all at once, automation will pay for itself quickly. We also have an application checklist, which covers the specifics of the job to be automated, such as space constraints, cycle time, how parts would be presented to a robot and how the robot would pick the parts. 
 
Which plastics markets are growing fast in their use of automation and robotics?
Patterson: A market in the Southeast area of the US that has grown fast is PVC pipe and fittings for construction and building, which is a high-volume application. There is more traditional six-axis automation than you might expect in those plants, with significant part variation.
 
Another area where we see strong adoption of automation is in medical devices, like plastic syringes, pipettes and IV bags. Cosmetics also have many high-volume products, and for higher-end consumer goods with products that are expensive, the makers want a very high-quality part, and they use robots to make those parts.
 
Sansoucy: When we look at in-mold labeling, medical parts have become a new frontier for IML. With the medical industry, the ROI can be faster because their product cost is higher and the risk of bad product is that much higher.
 
But I don’t think there is an industry that is immune to automation. We see it everywhere, from automotive and medical to consumer to just about anything being produced. The shoot-and-ship product market is still there, but we don’t see a lot of that growing through automation.
 
What factors limit the addition of automation to a production line?
Patterson: The first limitation is floor space. Often, when I visit a plant and scan the operations floor for adding automation, I see a facility that is wall-to-wall with injection molds and other equipment. There’s barely room to walk down the aisle, let alone get a forklift or pallet jack through.
 
The other limiting factor is what the production looks like in terms of mix and how many part changeovers occur. It is possible to do automation with low-mix, high-volume parts because robots are easy to program and set up if there are similarities between different products. The end-of-arm tooling may not need to change much; instead, it’s a matter of changing the robot’s position to pick up items. Robots execute programmed motions without adaptive reasoning unless they are enhanced with advanced controls.
 
Sansoucy: I think part complexity is a driver. Until someone designs automation, they don’t realize how complex it is, or what a human can do with hands and fingers to manipulate parts. If there are some very delicate inserts that need placing, the human hand is phenomenal at that. It has touch, and it can sense pressure. To mimic that with automation, even if the volume is high, is very difficult to do. It often requires product design changes and mold design changes. So, while volume is a factor, we also need to look at the intricacy of the part handling. We still haven’t designed anything as sensitive as the human hand, and every nuance of the wrist calls for another servo motor.
 
What are the limitations in automating in-mold labeling to eliminate secondary processes? 
Sansoucy: It’s volume. If you look at in-mold applicator placements, it’s volume vs. the price of the product.
 
If it is a very intricate product, like an electronic component that may have in-molded electronics, the inserts the product contains are very expensive. It is crucial to make sure that the insert is in the right position, because each placement that isn’t perfect will result in waste. That factor may justify automation.
 
With consumer products, it is strictly a volume game. In that case, the question is whether to print or to use in-mold labeling. That requires some further consideration, such as the shape of the product. A round shape is easy to print; a square shape is not. Even though a customer could justify smart automation for almost anything, volume and label complexity still are decision drivers.
 
IML once was reserved for premium brands. In a US grocery store 20 years ago, in-mold labels were rare. Fifteen years ago, IML was only for the top-end brands, the boutique brands. Now, everyone, even store brands, has in-mold labels. With everyone wanting to get into the IML space, prices have come down, and the design of IML systems is much more flexible.
 
When a customer adds robotics or other automation to a molding or decorating process, what type of training is involved?
Patterson: Many end users buy their automation from integrators. But, there are some that integrate the automation and robots themselves. So, we have two different types of customers.
 
A customer buying from an integrator typically gets a turnkey cell with a human-machine interface, or a dumbed-down version that has everything on the control panel for the robot. When the system is delivered, the integrator will give one to two days of training, or up to a week, depending upon the contract, to train operators across different shifts at the plant to operate a turnkey automated cell. Getting an operator up to speed usually takes a couple of days. That covers things like how to cycle start, cycle stop or reset home and how to adjust for different errors.
 
We also have customers that buy robots directly from us. They do their own integration and their own tooling; everything is done in-house. For them, there is a little higher learning curve. Those customers typically go to a four-day training course at one of our training sites, and the learning curve usually is three to six months from buying the robot to full integration. Once the initial team completes the integration and is ready to hand off to internal operators, they typically can train the operators in a couple of days to cover the basics of operating the cell, or by sending the higher-level technicians to our four-day training course.

Sansoucy: The big challenge for automation providers is making this equipment as easy as possible to use for semi-skilled operators, while at the same time giving the customer’s advanced users the ability to do the programming. At Arburg, we have the same controls for machines and for robots, and the system kind of teaches operators itself. But in general, if you look at it from a whole industry perspective, this also is a challenge we face – getting skilled labor now.

This is an opportunity for our industry to work with community colleges and high schools. We even have an outreach where we reach out to the parents of elementary school kids. That sounds crazy, but those parents can encourage their kids to look at this industry as an alternative career path; to work in an air-conditioned environment that pays good money. It is important for the industry to do outreach because if we don’t get skilled people, we can design the most advanced system in the world that saves all this money, but nobody knows how to run it.
 
With this automation, the more advanced it gets, the more we need skilled people. Customers cannot only rely on the equipment suppliers. If it is a Sunday morning and the automation goes down, the production schedule still stands, so there must be somebody who knows how to fix it.
 
What do you see for the future of automation and robotics? And is artificial intelligence (AI) going to influence that?
Sansoucy: “Artificial intelligence” is probably my least favorite phrase. I don’t think people know what AI really is. But this idea of creating a database of information to teach a robot or a machine to do something, I think that’s going to be the future.
 
One of the big technology areas where we see an opportunity is in vision-guided robots. Instead of having a bowl feeder and a very expensive mechanism that orients a part, the plant can use a vision-guided robot. The robot can make last-minute adjustments when it picks something up. With vision-guided robots, changing production from part A to part B to part C doesn’t require setting up different handling systems. The robot chooses how and where to pick up a part.
 
In another type of technology, there are companies that take mold designs from wherever they can obtain them and put them in a big database. Customers can specify a product and get a mold design in five or 10 minutes. The customer then can take that information, feed it into the molding machine and also feed it into the automation to prep the robot, the end-of-arm tooling and the whole process.
 
We also now have technology that accepts specs for the plastic material, the wall thickness and other info, and it figures out a sort of process. Language models could hold 1,000 mold designs, with the ability to make design changes instantly. That’s technology that’s coming to our industry, too.
 
Patterson: On the robotics side, I like the phrase “machine learning” better, because that is really what it is – training on a bunch of images and data sets to create things, whether it is vision guidance or path generation. One of the two biggest advances we see on the industrial robotics and the collaborative robotics side is in vision guidance. The idea is to eliminate the need for training on images of everything. A third-party camera company already has trained millions of images of all these different types of products. A customer just needs to train the system on its own specific part to describe and identify it. The vision system filters out and tries to help so the customer can throw anything in front of the system, and it knows what it is.
 
We’re starting to see more of that in the startup space. We will see if it makes it into mainstream manufacturing and production, because typically that is the challenge. Startups have cool ideas, but it is scaling success or failure that determines whether an innovation is viable for the rest of the industry.
 
The second big advance we see is automatic path generation for mobile robots, which is a tricky project. This gets into kinematics and the ins and outs of robot motion. We are seeing people use AI for motion-path planning for everything from bin picking to standardized applications. For example, AI could take a CAD model and be prompted to create a general plot for what needs to occur, and AI figures out all the motion. Those are in the early stages right now, but they still are very exciting technologies. The limitation with the current technology is that every manufacturer we visit, even if they have the same type of application, each one is somehow different. It will take collecting many data sets for that to be viable for our industry.
 
Conclusion
Automation via robotics and versatile new technologies offers tools to plastics decorators to improve product quality, cut costs, slash waste, speed production and ease workforce demands. From single-robot implementations to fully integrated turnkey systems, suppliers are positioned to support adoption across a range of production environments. However, as technology continues to adapt, operators continue to play an essential role in effective application of these automation systems.

Arburg is an injection molding machine supplier based in Germany with more than 100 years of experience. The company also designs and builds turnkey automation systems. Sansoucy has been in the industry for 30 years, working in engineering and operations, and now in sales
at Arburg.
 
The core of FANUC America Corporation’s product portfolio is servo motor technology, enabling applications across robotics, CNC machines and molding equipment. Patterson is a district manager, responsible for the state of North Carolina.

For more information, visit www.arburg.com and www.fanucamerica.com.