Robots Embrace the Daily Grind of Abrasive Work
POSTED 10/25/2019 | By: Winn Hardin, Contributing Editor
There should be a lot of praise for today’s robots, not the least of which is that robots never get rubbed the wrong way or complain about tough working conditions, grinding work environments, or abrasive bosses. This is true particularly in material removal and finishing applications, such as grinding and sanding, where their human counterparts face safety risks on multiple fronts.
“When it comes to grinding and sanding applications, a major driver to automate the process is worker safety,” explains Max Falcone, business development manager at PushCorp, Inc. “Especially when you look at an application like random orbital sanding, which is a high-vibration application. Many customers have stated to us during visits that applications such as random orbital sanding and grinding have contributed to their work loss days. Nerve damage and carpal tunnel syndrome have been mentioned numerous times as a driver to automate these processes.”
Deburring and Deflashing Versus Sanding and Polishing: Understanding Material Terminology
Material removal applications vary significantly, and so do their automated solutions. Knowing the terms can help guide potential customers to the best sources for information, technology, and service.
Deburring is about removing extraneous material from parts manufactured by subtractive processes, such as cutting, grinding, or milling, according to 3M’s Scott Barnett. Removing large burrs is often a two-step process. For some deburring applications, Barnett suggests a chamfer, or angled cut can be used. For deburring or deflashing applications where material stresses should be avoided, Barnett recommends a radius, where the force and position of the rotating surface of a compliant abrasive are actively controlled through electrical or pneumatic means.
Deflashing is used to remove material from parts made by additive processes, such as injection molding of plastic. Depending on the material, size, and other application requirements, flash material can be removed by any number of rotational media, from orbital sanders to grinders and other hard tools.
Sanding is a well-known surface finishing technique used in a multitude of ways, such as refining the finish of manufactured stone for sinks and tubs. However, understanding how media interact with different materials, including the details of wear, heat, and finish, can require an expert, says Britta Iwen, global marketing manager at 3M.
Polishing applications appear to be more cosmetically-driven than grinding and dimensioning applications. But 3M’s Carl Doeksen says to beware of that common wisdom. “Polishing might appear to always be a cosmetic thing, but that’s not always the case. A 6ra finish on a tungsten carbide-coated actuator is critically important from a functional/structural integrity standpoint. The surface roughness properties, achieved through precise polishing, allow the hydraulics to operate under extreme conditions and interact properly with the gaskets.” Polishing might also require a compound, coolant or other dispensed material to assure no thermal damage during the production process.
In truth, a single work cell can require one, some, or all of these automated processes. “Both sanding and grinding are multistep processes,” explains PushCorp’s Max Falcone. For sanding, you start by knocking down high spots and then go to a finer grit. Polishing is the last step. With grinding or deflashing, you might use a mill or a diamond abrasive, cut or snag the flashing out, and then finish with a grinding wheel.
Sanding and grinding applications can also adversely affect air quality, even when used in conjunction with industrial vacuums and air filtration systems. “Sanding applications, in particular, can create very dusty environments,” adds Louis Bergeron, integration coach at Robotiq. “As a result, it can be very hard to find good workers and to keep them. And until recently, there were automation solutions for only a few areas of the finishing operation.”
Workforce challenges are accelerating this trend. “We’ve been helping customers use abrasives in robotics for 30 years,” says Scott Barnett, abrasives application engineering manager at 3M, “But it’s really taken off in the past 3 years: our customers just can’t find workers.”
While robot sanding, grinding, deflashing, and deburring have been around for 30 years or more, market adoption has been slow, mainly due to the complexity of automating the task, leading to high costs. According to Robotiq’s Bergeron, most metalworking shops aren’t laid out with robotics in mind. “So when you want to automate the process, you have to start from scratch,” he says. “Many times, everything has to be changed. The way you handle the parts. The floor plan has to change if you’re adding industrial robots. It’s a big step to go from manual to an automated solution. . . . It takes a lot of skill to [automate surface finishing] well. The process has a lot of variables that affect the end result. You need to handle the robot speed, the force applied, the media, the kind of media you have, media wearing and changeout, the number of passes to achieve a specified result, the angle of the tooling, part position, and more.”
So why even try to automate material removing and finishing operations? Ask Ron Potter, director of robotics technology for Factory Automation Systems (FAS) and his client MTI Baths, maker of solid stone tubs and sinks. “A tub that used to take them four to five hours now takes less than two hours for a more than 150% increase in productivity,” Potter says.
Part of that productivity increase comes from the robot’s greater strength compared to human sanders, grinders, and metalworkers. “One of the things you can do with automation is you can run larger, heavier, stronger types of media than, say, a manual grinding wheel,” adds PushCorp’s Falcone. “I recently visited a customer who was running an 8-inch fiber disk for grinding. By moving to a robot, they moved to a 24-inch diamond saw, which reduces heat put into the part and eliminated warpage. So when you run a robot with larger media, you’re getting productivity improvement along with safer operations while enabling applications that weren’t possible before. There’s no way a manual operator could safely operate a 24-inch disk in all of the different orientations needed, for example. Overall the whole process is now better.”
Finding a Good Application
Everybody knows you can teach a robot to weld or do pick-and-place, but very few people broadly feel that you can teach a robot to do grinding or polishing operations,” notes Carl Doeksen, global robotics and automation leader at 3M. “There’s a perception in the industry that grinding and sanding — especially when you’re doing high-end finishes — that it’s a black art, that [the business] only has two or three people in the plant that really know how to finish these parts. [These workers] are in charge of the quality. But that’s changing. Today, any polishing, grinding, or finishing application can be automated.”
Doekson continues, “However, the question remains: Can an integrator make money on the job? That’s a separate question. It may not be economically practical to do it. And that’s certainly where we help integrators and industry to make those tough calls.”
“Sometimes a client decides on a robot, and they want to use it on 100% of their parts,” says Robotiq’s Bergeron. “We have to explain that you spend 20% of the money to handle 80% of the parts. Going to 100% of the parts can ruin your margin.”
Robot Programming Advances
Programming a robot’s movements — regardless of whether the application requires deflashing, deburring, sanding, or polishing — can account for a significant portion of the work cell’s cost, especially if the part is contoured or complex.
Material removal robots are typically programmed in one of two ways: off-line programming, possibly automated based on a CAD model of the part; and teach pendant or teach point programming. “We see off-line programming improving and getting better and better, but in many cases, it requires extensive programming skills and cannot suitably account for part variation,” says 3M’s Doeksen. “As off-line programming [leveraging CAD models] improves, we see the masses being able to do more of this type of application. If your application requires you to teach an extensive number of points on a grinding job, that can be prohibitive.”
While off-line programming based on CAD models has advanced to the point that robots can be using a milling tool to create celebrity statues and Stanley Cup reproductions, Robotiq is among the companies working to simplify the teach pendant programming process.
Robotiq’s new Universal Robots–based sanding kit reduces the time required to program an application for the robotic sanding of contoured surfaces from hours to minutes. “All of these applications need to be tuned because there are a lot of factors to program the robot path,” says Bergeron. “By moving the robot to only six points that define the contoured surface and telling the system the number of passes and interval between passes, we can generate a sanding robot program in minutes. If you’re not satisfied with the results, you can increase the number of passes, for example, and perform another trial. To do that with teach pendant programming, you’d have to use a ruler, intervals, find points, and ID each angle from that point. If you want to make a change, start all over again.
Force Compliance for Precision Work
In addition to programming, determining the correct amount of force to apply to the material removal tool is an important consideration for any robotic work cell.
For example, when MTI Baths asked Factory Automation Systems to automate its stone tub and sink sanding operations, Potter opted to use the low-profile FANUC force sensor on the end of FANUC’s M-710iC/45M robot. “When we analyzed the application, speed wasn’t critical, but reach was,” says Potter. “The robot we chose had 2,600 millimeters of reach, while the low-profile force sensor made it easier to provide active force compliance control in sinks and other compact internal surface areas. Using a larger active compliance tool on the end of the robot would have been too tight for sanding sinks and smaller products.”
To program the application, FAS scanned the tub and imported the 3D model into FANUC’s RoboGuide robot programming software. They then broke the tub down into nine separate surfaces, both inside and outside the tub, each with its own programming segment. it was determined that two passes with 80-grit paper, followed by another pass with 180 grit, resulted in a more uniform finish than manual sanders could achieve because they have to constantly move around the tub. “If you’re moving around the object, you inevitably miss areas and the consistency suffers,” said Potter. To accommodate changes in part positioning and part variations, the robot starts the process by touching the tub in eight programmed points to determine its position in the workspace.
According to 3M’s Barnett, flat parts make for much simpler automated solutions. “Often you can use position-based programming and not worry about force control beyond passive compliance, such as a foam pad [to accommodate part variation]. . . . Variability in parts is one of the biggest challenges for robotic applications.” Today, in-line scanning systems, used to measure actual parts in 3D to accommodate part variation, are finding their way into commercial practice when applications can support the additional cost.
“Passive [force] compliance is great,” adds PushCorp’s Falcone. “It’s a significant cost saving where a passive unit makes sense. They’re good for long planar moves, like the top of a table and the edge. But if you’re sanding a gas tank on a motorcycle, you need active compliance because the robot needs to move unhindered and control the force at all times. Active force compliance is like cruise control on a car. You set it for 50 mph, and cruise control tries to keep it there regardless if you’re going uphill or down. Active force compliance allows you to set it for 5 pounds of force, for example, regardless of whether you’re sanding the top of the table or the underside the force will stay consistent.
“When you look at other approaches, such as through-arm torque sensors and other collaborative solutions, they’re trying to control hundreds of pounds through six axes of motion while maintaining only 5 pounds of force,” continues Falcone. “Having the force compliance unit on the end of the robot makes for much faster response times, much more nimble solutions.”
What’s Next: CAD to Completion
Despite the challenges of robotic material removal applications, there are good reasons to keep an eye on this market. According to 3M’s Doeksen, the big robot applications, such as welding, assembly, handling, and painting, are saturated markets — but there are huge opportunities for customers and integrators in the white space of material removal. “In some areas, it comes down to: Do you want to compete against 10 companies for a welding application or be the only show in town for material removal?”
And the application base may get a huge boost in the arm in the near future, continues Doekson. “The most exciting area for these applications to me is 3D-printed parts and additive manufacturing, which is another industry that’s accelerating like crazy. When they get into metal and mission-critical parts, the post-processing of those parts is the process constraint. Typically, if you’re going to print a metal part that’s going into a jet engine, there’s a ton of post-processing that needs to be done. In the future, you can visualize an additive manufacturing cell, where the robot takes the output and post-processes it in place. That’s what’s coming. And it’s going to create some new challenges and opportunities for all of us.