Edge. Cloud. Digital twin. AI. AR. VR. Cobots. Once they were buzzwords. Now, they are becoming technical realities in mid- to large-scale manufacturing plants in North America. Timelines for adoption will vary according to industry sector, but a recent study indicated that 76 percent of manufacturers have a smart factory initiative in the works.
When implemented correctly, these initiatives will transform our plants—making the production of everything from prepared foods to jetliners more efficient, more consistent, and more interconnected. Automation enables a level of accuracy and productivity beyond human capabilities—even in environments that would be considered unsafe for humans.
The key phrase above is “when implemented correctly.” There are challenges and potential pitfalls. Manufacturing Engineering spoke with several industry experts to get their views on where we are today, what tomorrow could look like, and what could cause us to stumble.
When asked about the current level of automation in manufacturing plants, our experts were hesitant to generalize about manufacturing as a whole, choosing instead to offer observations about specific industry sectors.
For example, automotive and aerospace are heavily automated, according to Claude Dinsmoor, general manager of product development for FANUC America Corp., Rochester Hills, Mich. “There are many production steps in the manufacturing of an automobile or SUV that are so intricately automated and streamlined that it would be impossible for humans to step back in and do them today,” Dinsmoor said.
David Suica, president of Fastems LLC, West Chester Township, Ohio, sees considerable implementation of automation tools at the machine or cell level. “Automation of specific machines, or even cells or lines, is prevalent,” Suica said. “But most of these are islands of automation. The ability to see a factory in a digital format, so that true productivity can be visualized, is still in the future.”
According to Andy Joseph, director of product development, and John Lytle, application engineering manager for Promess Inc., Brighton, Mich., the degree of automation in assembly processes is less than on the machining side. “We do a lot of work in automotive and the automation of parts assembly is a mixed bag,” Lytle said. “On the other side,” Joseph added, “smaller manufacturers are looking into automation technologies that automotive has deployed for years. We’re seeing a big move away from hydraulics to servo-controlled actuators that can be part of the automation strategy of the future.”
At Praemo, Kitchener, Ontario, the principals see two areas of automation with very different trajectories. Andy Henderson, chief technology officer, said that physical automation (automation of physical and repetitive tasks) has been occurring over the last 30-40 years and there are examples in virtually every manufacturing plant. Decision-making automation (automation of human thought processes and insights), on the other hand, is very new. “Specific machines can perform mundane and repetitive decision-making today,” said Alex Kelly, director of solutions architecture, “but their ability to learn from prior events so they can adapt to new circumstance is not there yet. That is the next level—taking action on data captured in an automated manner.”
What are some of the newer technologies that set the stage for next-level automation?
“The acceptance of connectivity and data exchange standards that work within Industry 4.0 and Industrial Internet of Things (IIoT) initiatives are fundamental and necessary,” said Artur Gugulski, Midwest regional CNC manager, Fagor Automation Corp., Elk Grove Village, Illinois. “This aids manufacturers and suppliers alike in using open communications protocols and standards.”
“In the mold and die industry, for example, we are seeing flexible manufacturing systems that take the time down from seven weeks to seven days,” said Dan Zeman, automation manager for MC Machinery Systems, Elk Grove Village, Illinois, a subsidiary of Mitsubishi. “A key technology to enable this is work management software, which coordinates robots and machines by processing programs, data, and remotely starting the machine to allow for long, untended operation. The other thing that is critical to next-level automation is the technologies surrounding palletized tools, fixtures, and workholding. With palletization, a robot can be employed to pick up a tool and place it in the chuck very easily and accurately. You can go from milling to inspection to the sinker EDM in one continuous process with the robot transferring [the workpiece] from machine to machine. The operation can be scheduled, prioritized, and monitored by management software to efficiently sequence multiple steps.”
Robotics manufacturer FANUC is looking at several key technologies that augment human performance, rather than replace it. According to Dinsmoor, “Some of the exciting technologies include machine vision (3D vision), tactile and force sensing, and power-and-force control, which are essential for cobots (collaborative robots that work in close proximity to humans). Also of high value is simulation technology at the design stage, which helps us build a virtual solution before committing it to steel. With simulation, manufacturers can reduce their design time and safely evaluate up to a dozen ‘design candidates’ to find the best automation solution.”
According to Paul Boris, executive vice president for Praemo, if we think of next-level automation as equipping machines to start making decisions on their own behalf, that requires a level of intelligence in the machine itself. “To take advantage of the different algorithms, and the methods for deploying those algorithms, we rely heavily on artificial intelligence (AI),” he said. “AI is the foundation that runs across a lot of other technologies important to us.”
One of the more interesting technologies Promess sees is LiDAR (Light Detection and Ranging). “Its primary characteristic is its high spatial resolution when measuring 3D objects,” said Joseph. “LiDAR can be employed in a variety of manufacturing scenarios, including operator safety, robotic parts handling, guidance systems, and machine learning for optimizing assembly processes.”
Zeman said the next advance MC Machinery is looking for is in the area of design. “Manual drafting gave way to doing everything on a PC and plotting on paper. Plotters have now disappeared as everything goes electronic. Eventually, instead of hitting the ‘print’ button, we’ll hit the ‘build’ button and the design will go straight to manufacturing.”
Other new technologies mentioned by our experts include edge computing (processing at the source to speed up transfer rates and response times while reducing the load on a factory’s bandwidth); native integration with the cloud; technologies to predict/prevent asset failures; augmented reality (AR) for assembly and maintenance; and virtual reality (VR) for training and simulation.
“To achieve next-level automation, several things must be happening at the same time,” Suica of Fastems said. “VR technology will be a big advantage as it allows a multi-faceted view of your factory in digital form. In fact, seeing a factory in digital format (i.e. digital twin) will bring productivity improvements. Digital visibility allows managers and engineers to instantly see which machines are not only on, but which machines are cutting, which is the true measure of operational efficiency. It also shows the flows between machines—where the disturbances or problems are.”
With software, the tendency is to keep adding capabilities. “This is fine,” Dinsmoor of FANUC pointed out, “as long as it doesn’t add layers to the overall architecture. By integrating more functionality into our robot, we actually reduce the layers of software and hardware required. Open protocols are essential for this to happen on a large scale.”
According to MC Machinery’s Zeman, the benefits of automation come from combining multiple process steps up and downstream. “This information must flow seamlessly,” he said. He spoke to the collaborative effort between manufacturers who are embracing open standards for data collection and file transfer. “The net result is that a customer can integrate CAD/CAM, measurement system, milling, and EDM machine easily and obtain excellent results,” Zeman said.
Lytle of Promess cited the advances made in machine-to-machine communications via Open Platform Communications with Uniform Architecture (OPC UA) and Fieldbus technologies for real-time distributed control. “Our customers want solutions that allow them to do the integration and re-programming internally,” he said. “Machine libraries that provide building blocks with little programming knowledge required, and things like Open PLC, are making this a reality. However, I should add that we’re at the very beginning stages here.”
Suica’s evaluation of the problem with islands of automation is the existence of “little bits of software that don’t talk to one another.” He equates it to sitting down in the evening to watch TV and having four different remotes in front of you—one for the receiver, one for the DVR, one for the TV, and another for the streaming device. “Just like you need a universal remote at home, your factory needs a universal way to collect data and use it to achieve a task,” he said. “This is getting easier by the day due to open platforms and architectures and databases, but it still must be planned for.”
Manufacturers hoping to implement an AI-driven analytical engine in the future can start today by unlocking valuable insights without having to prepare all the existing data in a special way. Henderson of Praemo explained that when employing traditional approaches, there is a requirement that key data need to be located, cleaned, categorized, and harmonized in order to become the foundation for decision-making.
“Unfortunately, even within the same company, you will find each manufacturing site doing its own thing—different databases, different ERP systems, different tags, different names,” he said. “Instead of waiting to complete data standardization, we connect to the data the way it exists, and where it exists, today. Start quick and get some small wins. Not only does this get things started faster, it helps companies address the people and cultural issues that often accompany large-scale automation projects, without the inherent risks or fears.”
Next-generation robots are not only much easier to program, but also easier to use, with capabilities like voice and image recognition to recreate complex human tasks. Collaborative robots have emerged as a new and fast-growing technology that can satisfy the demand for automation and flexibility in manufacturing. They have the potential to assist workers doing tedious tasks and improve workplace safety.
According to Dinsmoor of FANUC, his company’s advanced robotics development work centers around answering this question: “What can tomorrow look like?” Emerging technologies such as vision, power and force control, AI, and machine learning are all part of the design palette. “Our intent is to make a robot easier to apply and easier to program or operate without having specialized skills,” he said. “Much like operating a smartphone that has amazing capabilities that are mostly transparent to the user, improvements in reach, speed, and payload are a continuous priority.”
FANUC has been embedding IIoT-compliant devices in its robots for the last five to seven years to enable remote monitoring toward the goal of zero downtime. “We have shops running 24/7 without human intervention,” Dinsmoor said, “so we see the reality.”
Our experts have differing views about the need to achieve lights-out operation. As Boris of Praemo put it, “Lights-out is a good vision for all manufacturers to strive for, but it has to stay rooted in reality. It is most beneficial for a plant that is manufacturing standard, repeatable products. Whenever you introduce variability into the equation, through customization or other factors, lights-out is much less achievable. We think it is important to layer-in intelligence, analyzing data from machines and processes as complexities arise, alerting teams before emerging issues result in performance or quality losses. So, perhaps the goal should be intelligent assistance rather than complete operational autonomy.”
Suica said that Fastems has several customers that are running lights-out (90 hours or more per week) consistently. “That requires several things to happen. First the machinery and equipment must be robust,” he said. “Then it’s a process of identifying the weakest links and eliminating them. In some factories, the weak link is tooling, in another it’s chip removal. Whatever is locking the machine out must be eliminated. But whether or not you achieve lights-out, identifying and eliminating the weak links will always lead to productivity gains.”
Gugulski of Fagor cited a list of “typical” requests from customers who are striving for lights-out. “Things like dual-channel, 10-axis systems; collision detection; machine safety; glass absolute scales; thermocouples for dynamic temperature compensation; tangential control; sophisticated probing; runout transformation; motion detection; and features to modify programs on-the-fly are all requirements for lights-out CNC machining,” he said. “This is doable, but complex. We’ve demonstrated the ability to interact with different components of the automation process, including bar code readers, thermocouples for temperature measurements, safety systems with cameras, radio-controlled probes, and top-of-the-line CAM systems for five-axis parts programming.”
The gentlemen at Promess hear less talk about lights-out from their assembly customers. “Human interaction is almost always required,” Joseph said. “In fact, cobots were developed to interact with humans.” According to Lytle, “The need from our customers is for flexible assembly systems—doing more at one station. Many plants are looking for our press to become a sixth axis for a robot, another end-of-arm application.”
Zeman of MC Machinery added, “For lights-out operations on qualified processes, for example in the aerospace and medical industries, we actively try to work with our customers early in the development process to ensure that the benefits of automation can be incorporated in the process. The first-article run has to include automation, as it cannot easily be added at a later date.”
The Industry 4.0 initiative requires basic digitization. Without digitizing every step, operational efficiency is being left on the table.
“One of the first things we recommend is an automation audit to look at all the process areas in the plant,” FANUC’s Dinsmoor said. “There may be a customer with very specific ideas about what should be automated and, while doing a walk-through, we find additional opportunities that are high value and easily achievable. For a plant just starting to automate, these early and easy wins are best. The successes feed on each other and the economic payback can often help fund the next opportunity.”
One way that Praemo tests the water with potential customers is to start fast with a static set of historical data, according to Henderson. “Using static data, our Razor engine can analyze and provide insights to plant management without disruption to existing production. We all came from operations backgrounds, so our emphasis is on the practical. We know from our experience that AI and machine learning will simplify, speed up, and reduce the cost of manufacturing.”
Technical feasibility and economic viability are necessary preconditions for automating a machine or line. Yet, they are far from being the only factors, according to Joseph of Promess. “Outside of the obvious technical issues, the key point remains: if you’re looking at automation today and have questions, who do you want to work with? Who will be there to support you as you take the journey?”
Connect With Us
