A single operator can sometimes tend multiple robots, each of which are tending multiple machines—it all adds up!
If “automation” is the constant drone you hear from practically everyone in metalworking these days, job shop owners might be the only people yelling “No!” Or at least “Wait!” How, they ask, can you cost-effectively automate low-volume, high-mix parts? Yet it’s not only doable but probably necessary.
As Steve Alexander, vice president of operations at Acieta LLC, an automation integrator with locations in Waukesha, Wis., and Council Bluffs, Iowa, put it, “The U.S. leads the world in manufacturing, but also has one of the highest manufacturing hourly labor costs. Job shops are not only competing with other local shops for business, they are also competing with other countries. Automation can lower their costs so they can better compete.”
He added that the demand for manufacturing capacity is outstripping the available labor supply and “job shops may find it difficult to compete with larger, more established machine shops” for ever-more-expensive talent. But a robot can tend multiple machines and, in some cases, a single operator can tend multiple robots, each of which are tending multiple machines.
“That frees up operators to perform secondary operations, like deburring parts, assembling or spending more time on quality checks,” said Alexander. To cite just two examples, Central United in Newburg, Wis., uses a single FANUC M-20iB robot to tend two lathes. Another Acieta customer in New Berlin, Wis., uses a single FANUC R-2000iB robot to tend four machines. “In both instances, there is only a single operator responsible for loading raw parts onto the conveyor or queue station for the robot to grab.”
Alexander added yet another compelling reason to automate: ergonomics. “Workplace injuries are a concern for any size shop. Something as simple as loading a part into a lathe chuck can quickly turn into a repetitive stress injury for an operator when that process is repeated hundreds of times a day. Between lost time and claims, these injuries cost companies billions of dollars each year. Properly integrated robots can remove a repetitive process from the operator, thus improving employee wellness.”
Options & Integrators
So far so good. But as John Lucier, automation manager for Methods Machine Tools Inc., Sudbury, Mass., explained, the simple goal of “making more parts more efficiently” can lead to very different approaches depending on the type of work being produced and the volumes involved.
For example, let’s say a medium-volume job shop needs to changeover a couple of times per week. “With proper planning a cell could be changed over on the manned shift, allowing for a nice amount of unattended production on the following shift.” And within this context, the required speed of the robot would depend in large part on the machining cycle time of the parts produced. The shorter the cycle time, the faster the robot needs to be and the less likely one robot can support several machines.
But shops with even lower volumes—even lot sizes of one—can also be good candidates for automation, Lucier explained. Except that “the most important consideration is reducing changeover time, or even totally eliminating setup time. People get caught up on cycle time and load/unload time, but if your volumes are so low that half your time is spent changing over, then attacking the changeover time, even at the expense of cycle time, may make sense.”
One way to “eliminate” setup is to automate it, and today’s robots can be programmed to change both their own end-of-arm tooling and the workholding in the machine. Another approach is to load common pallets from a carousel (or another type of stacker) regardless of the part being produced. This eliminates the need to change robot tooling or to program different load/unload moves for different parts.
The Methods PlusK system takes this concept a step farther and also loads toolholders from the carousel. Designed to support FANUC’s RoboDrill series machines, the PlusK’s carousel has five rings, each of which has 12 stations, for a total of 60 stations. Each station can hold either a part pallet or a toolholder and the robot can load both parts onto the RoboDrill’s worktable and tools into the machine’s integrated 21-tool turret.
But as Lucier pointed out, if you want to automate the production of 30 or 40 completely different parts, 21 tools may not be enough. “We commonly load parts in tiny five-axis vises from suppliers like Lang Technovation,” explained Lucier. “And we usually mix and match between parts and tools in the carousel. Let’s say one job requires five tools that differ from the tools used on the previous job. I can use the robot to load each tool as needed, or to load all five tools into the turret at once, much like an operator would set up a job.”
If these considerations are starting to make your head hurt, that’s understandable. Because while it’s gotten much easier to implement automation in recent years, it’s still difficult enough that most job shops would be wise to take advantage of integrators like Acieta and Methods. From Lucier’s perspective, programming the robot is the easy part. Things like dealing with the automatic workholding, chip removal in the work zone, and how best to present the parts to the robot are the bigger challenge. That’s where expert advice really pays off.
Automating Legacy Machines
Here’s an example of how that can happen. By mid-2017 Gary Kuzmin, owner of All Axis Machining in Dallas, Texas, was a textbook example of job shop frustration. He had more potential work than the available workforce could support and a shop full of high-end German five-axis machining centers, EDMs, and other technology the OEMs couldn’t economically automate with the kind of versatility that could get him out of this bind. His solution was to invest in a collaborative robot from Universal Robots (UR), Ann Arbor, Mich.
Like any job shop owner, Kuzmin’s top requirement was retaining “complete flexibility” in his operation. “We’ll make a part for two or three days and then we’re on to another. In a given month we could be making 30 to 40 different parts on one machine.” So he needed a robot that was easy to set up and change over. And he needed features like force-control grippers so the robot could adapt to different parts and the change from raw stock to a finished, thinner-walled part.
UR delivered on all counts. Its robots have a programming interface on a hand-held pendant. They operate safely side-by-side with humans because the robots automatically stop when contacting an operator (hence the term collaborative robot, or “cobot”). They’re easy to train and not intimidating.
Kuzmin explained that a key reason for the flexibility of UR’s solution is an open architecture that has allowed third parties to develop software, vision systems, end-effectors and other accessories that work with UR robots. UR facilitates the integration effort by providing a one-stop on-line showroom called UR+.
Even with that, Kuzmin found that there was no ready solution for programming the robot to perform sanding or deburring, two functions he wanted to automate because they are tough on humans, and especially older workers. So All Axis used UR’s free software development kit to create its own routines, which follow the curvature of the part based on the CAD model.
These plug-ins integrate the control of the application through the cobot’s own teach pendant. “Universal Robots has done a fantastic job of making this solution available to shops like ours,” said Kuzmin. “We’re not locked into a proprietary environment. Instead, we’re given a cookbook with an easy way to expand communication,” adding that All Axis uses protocols such as TCP/IP, Ethernet, Modbus and Profinet to connect cobots and machines.
“Another advantage is we don’t have to pay special license agreements—it all comes with the purchase of the UR cobot,” said Kuzmin. “We sign up on the UR website to become a developer, and we download the elements we need.”
In another example, All Axis decided it wanted the ability to move a cobot from machine to machine as quickly as possible. The challenge was creating a system that both reoriented the cobot to the next machine and secured it to the floor without drilling holes in the concrete. The solution was to epoxy a receptacle onto the concrete and attach an indexing pin.
“If the robot ever needs to be moved or serviced, you simply unlock and pull the pins and move the robot as needed,” said Kuzmin. “When you re-position the robot and lock down the pins, everything is still indexed. We also attached the presentation tray to the robot stand, so its position relative to the axis of the robot never changes.”
Within just a few months, Kuzmin and his team got so good at integrating cobots (they now have seven) and so excited about the benefits of automation, they created a separate company called All Axis Robotics. While Kuzmin said it’s relatively easy to automate a new machine by spending about $150,000 with the OEM, he sees tremendous potential in automating legacy machines, which is their focus.
To take one example of how automation helped the bottom line at All Axis Machining, one job that typically took six months to run finished in 2½ months by using the robot. “It was phenomenal. ROI was about seven to eight months,” said Kuzmin.
In perhaps its most impressive innovation, All Axis combined a UR robot with an off-the-shelf microscope and camera, plus artificial intelligence (AI) and deep learning, to create a system that detects burrs, rolled edges or any anomaly (such as missing hardware) on a part.
That’s important, because as Kuzmin explained, many burrs cannot be identified except under magnification and a human’s eyes will become fatigued after just 20-30 minutes of such inspection. But it’s also difficult to automate.
“Burrs are like snowflakes—no two are alike,” said Kuzmin. “Our system, which we call Aurora, uses AI to teach itself what a burr looks like. The more burrs it sees, the better it’s able to detect them. The system will identify the burr, classify it, and sort the part after it has rejected it due to the burr. It also records the inspection and the burr’s location so the operators are able to go back to efficiently correct the part and move it on.”
All Axis introduced Aurora in 2018 and Kuzmin says the single biggest response has been: “We’ve been looking for something like this for years!” An upcoming release of the system will include the ability to insert helical inserts (commonly called Heli-Coils) and hardware into parts as well.
Industrial Robots Fight Back!
As discussed, UR cobots are easy to set up and integrate with a variety of machines and
accessories. But industrial robot providers like FANUC and KUKA Robotics took notice of features like the intuitive hand-held teach pendant and introduced their own versions.
Method’s Lucier acknowledged that “at one time it was absolutely the case that collaborative robots were easier to program. But setting up an industrial robot is no longer really any different from setting up a collaborative robot.” Even the “grab it and drag it” setup function is available on traditional robots, though Lucier said this is best used for cases in which you simply want to set the part on a table or against a block.
“It’s perfect for something like teaching the robot to load a CMM because the machine just needs to know approximately where the part is in order to find the edge, and then go on to conduct the measurements,” he said. “But if I’m loading into a chuck or a vise, I have more control when I teach it in with the programming pendant.”
There are also a variety of accessories for industrial robots. For example, Lucier said, quick-change and force-control grippers are not unique to cobots. “Schunk has a line of force-sensing grippers that use servo motors, getting away from pneumatics, which are harder to switch. You can really dial in the electrics.”
The industrial robot folks also have solutions that allow you to quickly reposition the robot from machine to machine. In FANUC’s case, a camera on the robot scans three decals on the “new” machine to reorient itself automatically.
But in months of trying, Lucier said they can’t find a situation in which such a capability is useful. “Moving a robot from machine to machine entails more than simply moving the robot. Think of all the things you need to change: Maybe the end-of-arm tooling, the in-feed, the out-feed…” The better solution is one in which each machine has its own robot or a robot services several machines, perhaps with dual end effectors so it can switch between part types without intervention.
It’s also the case that integrators like Acieta and Methods can add traditional robots to legacy machines, and robots are priced about the same as cobots. So considering current offerings, the biggest advantage to cobots is their ability to function near humans without the space and expense of fencing.
The disadvantage is that they are slower than traditional industrial robots and are limited to lighter parts. UR’s largest model handles parts up to 24 lb (10 kg). Slower load/unload times aren’t significant on parts with long cycle times, but they can decrease output on medium- to high-volume jobs with shorter cycle times.
Kuzmin of Axis Machining doesn’t see the need for speed, “just consistency and reliability—especially on parts with a long cycle time. It’s better to have a cobot load parts at a reliable speed and then use it to deburr other parts while the machine runs. And if you don’t need the speed, why risk it?”
Lucier and Alexander caution against exaggerating the benefit of being able to work right next to the robot. For one thing, Lucier pointed out, you already have to shield the operator from the machining area. Alexander added that if a cobot is handling a part with sharp edges (as is commonly the case in machining), “it’s not collaborative or safe for an operator to work in its proximity anyway.”
Both Lucier and Alexander said combining “collaborative techniques” with industrial robots offered a useful compromise. “For those manufacturers seeking an open floor with minimal fencing, industrial robots have the ability to work in a fenceless environment by using area safety scanners,” explained Alexander. “And they can work in this fenceless environment without sacrificing speed. When an operator enters the first safety zone, the robot slows down. When the operator enters the second safety zone, the robot stops completely.”
On the other hand, said Lucier, since the safety zone for this approach is 3-4′, the robot would often be slowing down if fork trucks or other traffic passed by frequently.
As suggested earlier, one area that’s perfect for cobots is in tending a CMM. “CMM’s are intrinsically safe to begin with,” as Lucier put it. “There’s no guarding, except around moving components. They’re not fast. You can stand right next to them. And generally you’re not putting the part into a fixture, perhaps just holding it against a V-block or something similar. That’s where I’ve seen a collaborative robot shine.”
Both experts agree, as Alexander summed up, that “industrial robots are much more applicable than cobots in the majority of job shop and manufacturing applications.”
According to Greg Mercurio, president of Shop Floor Automations, La Mesa, Calif, “software is the most essential component in job shop automation.” He said direct numerical control (DNC) software (also known as distributed numerical control software or “drip feed” software) “allows for streamlined machine communication, as well as revision control for managing quality/scrap, and a means to organize thousands of programs.”
The main function of DNC software is to send machining programs directly to the machine, enabling a machine with a relatively low-grade control to handle a complex surface.
Mercurio and all the experts agree that some degree of machine monitoring is helpful “to increase uptime and machine capacity.” Mercurio added that Production Data Management software enables companies to go paperless and that “tool crib management is also a necessity. The best part is that these solutions can all integrate with CAD/CAM, ERP, MES, CMMS and more.”
On the other hand, he pointed out that smaller shops can greatly benefit from a simple hardware addition before spending thousands on software. For example, adding an RS232 to USB interface to older equipment makes drip-feeding programs easier.
“The hardware is easy to use, set up and maintain, so it is the least time-consuming automation
option for a smaller shop. Older RS232 machines can use a USB drive like newer CNC machines that have that function built in from the factory. It’s also a more affordable way for them to modernize their shop floor without replacing equipment altogether. It also allows for affordable memory upgrades,” added Mercurio.
Getting back to robots, Alexander said there are numerous software packages offered by robot OEMs and third parties that enable robot programming off-line on a PC while the robot keeps working. “These software packages allow you to load a CAD file and then give you the freedom to ‘draw’ your robot path using a stylus or a mouse,” explained Alexander. “The software will then convert your drawn path into robot code that can be loaded directly into the robot.”
It’s fitting that job shop owner Kuzmin gets the last word: “I wanted to improve productivity. I did it by adding robots. And I’m very pleased with what we’ve accomplished. While doing it I said we ought to make this available to everybody out there, because that’s what’s going to make manufacturing in our country much more competitive.”