Skip to content

Machining with Robots

Pat Waurzyniak
By Patrick Waurzyniak Contributing Editor, SME Media

Robotic machining technology has advanced to where it poses a serious alternative to metalcutting applications on more traditional machining centers. With the latest robotics equipment and related software, automation suppliers and robotic system integrators are gaining some traction using robots in many material-removal applications previously done only with machine tools.

Rigidity and accuracy remain the major obstacles to widespread use of robots for metal material-removal jobs, making robotic milling of harder metals difficult. However, robots have been widely deployed in material-removal applications, including deburring, deflashing, trimming, polishing, and grinding. Waterjet and laser-cutting processes are also being robotically controlled in material-removal applications, while robotic milling of softer materials, including foam, plastics, wood, sand castings, and aluminum.

A parallel-link F-200iB Fanuc robot offers the rigidity required for metalcutting applications.

Robots offer the advantage of a very large working envelope compared with standard machine tools, says Robert Gian, professor of mechanical engineering at the Lee-Ming Institute Technology in Taipei. “Why use a robot? They can run 24 hours, they can do occasional material handling, and can be used for painting, deburring, and assembly,” notes Gian, who recently gave a presentation on robot machining at the Delcam plc (Birmingham, UK) Asian user group conference in Korea.

Employing Kuka Roboter GmbH (Augsburg, Germany) KR60 robots equipped with Kuka’s CAMRob and Delcam’s PowerMILL CAM software, Gian says manufacturers can use the robots’ turning range of ±185° and its large working envelope (about 27.24 m3) to machine very large workpieces, and mill shaped parts from softer materials. Robots offer a high degree of flexibility with up to eight or more axes, and repeatability of ±0.15 mm that can be utilized in a variety of applications, says Gian.

With its seven-axis Kuka robot, Garner Holt Productions mills animated characters used in theme parks, casinos, and museums.

For aerospace and automotive applications in robotic trimming, drilling, and machining, robots are becoming more common, even with articulated arms’ inherent limits on rigidity. Some shops are using robots to machine big sand castings, instead of pattern-making for casting very large metal parts. “We have customers that machine a lot of sand for the foundry industry; instead of making a pattern, you just cut the sand directly,” notes Tom Bentley, president, Programming Plus Inc. (PPI, New Berlin,WI), a reseller of robots from Kuka Robotics Corp. (Clinton Township, MI) and Delcam software. “They’re a humongous money-saver, if you do lower-volume kinds of things, typically for steel foundries that make larger one-off components. Especially if you make big parts, if you’re making 30,000 lb [13,500-kg] castings, that’s a $40,000 pattern to make one part. You cut the sand directly, you don’t make any patterns. It’s pretty effective.”

Robotic trimming of woods, urethanes, and sand castings is very common, Bentley says. “They’re used in trimming, and they’re used in a lot of architectural sculpture work,” he adds, noting the case of customer Garner Holt Productions (San Bernardino, CA), which used Kuka robots in an RMC100, a seven-axis Robotic Milling Cell (RMC) from PPI’s sister company, Robotic Solutions Inc. (New Berlin, WI) to create realistic, life-sized animated characters for theme parks, casinos, and museums.

The barriers for robots moving into milling is robotic systems’ accuracy, Bentley notes. “If you think about it, the tip of the tool is basically stuck on an end of an arm which is a hundred inches long with a couple of joints in the middle, so it is amazing what they do. It’s a tolerancing issue. Considering the fact that you’re on a 90–100″ [2.3–2.5-m] arm, they don’t really have the rigidity to do it. As soon as you’re talking aluminum, users are looking for a little different set of tolerances than we can achieve. If you’re talking about cutting plane skin, we can trim thinner-gauge aluminum components.”

MotoSimEG-VRC software from Motoman allows users to simulate robotic workcells prior to finalizing programming for material-removal applications.

Repeatability achieved with Kuka robots is the best he has tested, notes Bentley, of the systems’ ±0.15-mm repeatability. “Assume you were to draw a line with a Kuka robot used for trimming aluminum, like a 100 LE—a good-sized robot, with about 11.5′ [3.5 m] of reach. If you programmed a line move from two points that were 6′ [1.8 m] apart and take a Faro tracker device and track it as it moves, mapping the points, that line would be straight within about 0.004” [0.1016 mm ], and in the robot world, that’sphenomenal. Machine tool world—you start looking to replace that.

“Once you move to metals, everything gets harder. We have robots deflashing aluminum components. They make a part about every 2 min, 30 sec, and we use a vision camera, find the cast features, and deflash them. If you’re talking about flash, that’s anywhere up to 1/8″ [3.18-mm] thick, 0.500–0.600″ wide [12.7–15.24], taking a 5/16″ cutter, and just cut it to size on the first pass.”

Cutting harder metals with robots requires much more rigidity than is typically offered in a standard robot. “There’s basically one rule of thumb that I look at when it comes to using a robot in a machining application, and that is the machinability of the material that you’re cutting,” notes Virgil Wilson, senior engineer, material removal, Fanuc Robotics America Inc. (Rochester Hills, MI). “Softer material, where you don’t get as much reaction force causing robot deflection, is a candidate for robotic machining. Generally, the line for a serial-link robot’s going to be aluminum. We’ve actually installed a couple of systems where we use a robot to cut aluminum. We have one unique product that allows us to cut carbon steel—low-carbon steel—and that’s the F-200iB, which is a parallel-link robot. The big difference there is a parallel link is going to be a lot more rigid than a serial link.

“On a parallel-link robot, you have a base and a top plate, and all of the robot’s legs are tied together between those two,” Wilson says. “That essentially gives you all the rigidity that you can get, whereas the serial links are just a series of links of rotary axes tied together to create the arm. And that really is the crux of what holds serial robots back from doing all kinds of machining, because the robot is inherently going to deflect as you get into the tougher materials, and will not stay on the path that you want it to follow.”

The Fanuc parallel-link F-200iB robots offer a substantially more rigid platform that enables cutting low-carbon steel in automotive applications. “When it crosses over into cold-rolled steel, we’re going to move into the parallel link robot,” he adds. “We’ve cut up to 1/4” [6.4-mm] plate with it. For example, cutting a hole in a plate for bumpers is one thing we’ve done quite a bit with that unit, just machining or milling a hole in the bumper, or the frame that goes around the door.

“Serial links are by far the most common robot out there,” Wilson says. “The trade-off with the parallel link is that it has a much smaller work envelope compared to a serial link, where you get the huge volume, the robot can slip behind itself, and so on. The serial link has a much more confined space that you can work in.”

Refining robot accuracy is another goal robot builders are working to improve, notes Wilson, particularly for addressing the aerospace industry’s accuracy requirements. “The other issue besides rigidity is the accuracy of the robot,” Wilson says. “Most of the time, the milling type of applications are going to be programmed offline, then downloaded to the robot, and now you’re commanding the robot to go to a position—and that’s where the inherent accuracy of a robot comes into play. Fanuc has, for the last 15 years, been developing tools to improve that accuracy. We have a whole family of tools available to correct it, depending on the level of accuracy that’s required for the application.

“Most of it’s done with software and the main premise behind the software is to correct the robot, and make sure that it’s mastered correctly, and to account for all of the inaccuracies in the actual building of the robot,” he adds. “In other words, you might have a little bit of an error in the casting, where it’s machined, and we have options that correct for that inaccuracy to build a model that is representative of the actual robot. That all resides on the robot. Basically, what happens is you make a kinematic model of the robot, and that assumes that everything is perfect. In reality, it’s not so we have software options that we add that enable the robot to calibrate itself and figure out ‘this link is actually a few thousands longer than it should be,’ for example.”

Fanuc offers that capability in a software/hardware combination that customers can purchase, Wilson says. “That’s one level, and then there’s another where we use a Leica tracker to measure the robot in space, go through a bunch of different positions, and then we back-calculate all the kinematics, or what we call DH parameters.

“Aerospace is asking for some very tight tolerances in drilling and routering, and in general terms, because it’s still being developed, we’re essentially using another set of encoders in our serial-link robots,” Wilson notes. “Instead of measuring the revolution counts on the motor, it actually has a way of counting outside of that drivetrain; in other words, it’ll absolutely measure the correct angle of each joint, and it essentially just bypasses all the backlash and all the lost motion. Mainly what we’ve targeted right now are the serial links, because that type is very attractive because of the work envelope, and typically that’s what you’re going to see in aerospace—they want large work envelopes versus a small envelope.”

Accuracies obtainable vary by application, but with robots using the newer encoder technology, users can achieve the accuracy requirements that aerospace builders are asking for, which ranges from ±0.3 mm to ±0.5 mm, Wilson says. Customers include larger aerospace OEMs, mostly in drilling and riveting applications, he adds. “Although we have seen a fair amount of requests for routering, and we kind of see that as the next step. Right now, we’re focused on position accuracy. Routering would involve a path-accuracy issue, and that’s the next evolution that we have to make.”

Robots today are being widely deployed in material-removal applications, notes Greg Garmann, software and controls technology leader, Motoman Inc. (West Carrollton, OH). “There’s a multitude of different projects. Primarily, we’re not using robots to gouge out big chunks of metal. We’re using them for finishing and polishing, but we’re also doing things like deburring, grinding, brushing, which are important in metalworking.

“Rather than having a very large CNC-style machine, you can also move in a capital-equipment cost-effective direction. We’ve done things from using a machining process, to doing operations like deburring. We’ve also used that same kind of process in waterjet cutting and even milling of lighter materials for mold creation.”

With software including Pointers in Porter with GCode Converter, Motoman makes it straightforward for users to program robots in a way similar to machine tool programming. “We’ve been involved in applications like waterjet cutting where customers use our Gcode converting product, which makes our robot look like a CNC-style machine. What we’re doing is, we’re getting information from CAD/CAM software that creates the G code, and we can postprocess that into the robot language.”

Another software package from Motoman is its MotoSim EG-VRC package, which recently was updated with its Enhanced Graphics simulation, which now includes Virtual Robot Control (VRC). “We’ve incorporated more capabilities into our simulation tool to show exactly what would be on the teach pendant, and you can control the robot just as if you were using a teach pendant on a factory floor,” Garmann says. “You can use this tool off-line in your office, on a PC, to program your system. There are points where the virtual world and the real world may not match exactly, so we model the system as close as possible to the realworld conditions and touch up some points on the robot. Typically the simulations are within a millimeter.”

In material-removal applications, Motoman’s EH80 robot and, for larger payloads, its DX1350N model, can handle a variety of tasks. Systems installed with Motoman customers include deburring applications in aerospace. “We’ve completed a number of systems for aircraft engine components in the aerospace industry,” Garmann adds. “There’s a big demand in aerospace.”

Drilling holes in composites for airframes and wings using robots has been a key application for ABB with heavier-payload robot models including the ABB 7600 500-kg robot used to drill holes in composite, and even in titanium workpieces, says Josh Williamson, lab manager, Robot Automation, ABB Robotics North America (Auburn Hills, MI).

“We’ve done some drilling in the wings, fuselage, and even some aluminum skins in the past in the aerospace industry,” Williamson notes. “We’ve actually used the robot to do the drilling, so we’re using some heavy-payload arms to make sure the robot is rigid enough. You can get a little bit more accuracy from it, if you’re traveling with the tool in the negative Z direction.”

Material removal is one of the least-utilized with robotic applications, but it’s growing fast, he adds. “The reason for that is it’s a dangerous, dirty, repetitive motion that really lends itself to carpal tunnel,” he says.

With ABB’s Force Control for Machining, which Williamson says is often used in grinding, offers sensors in the articulated arm and software controls that help control the amount of force used at the end-of-arm tooling. “The force-control machining applications use a transducer at the end of the arm that gives feedback directly through our axis computer, and the robot can sense forces at the end of the arm,” Williamson says.

“Say you need to polish a wing. We’ll set the speed and the force that the robot needs on the part. If there’s a variance on the part, or the part isn’t perfectly located in a fixture, then the robot will vary its path, but maintain the surface force, using the force-control machining. If you’re grinding off welds, the robot will slow down when it sees a certain amount of force on the tool. That allows you to keep your tool safe, your tools last longer, and you don’t have to worry about breaking your spindle or the machine you’re using.”

This article was first published in the November 2009 edition of Manufacturing Engineering magazine.

  • View All Articles
  • Connect With Us

Related Articles

Always Stay Informed

Receive the latest manufacturing news and technical information by subscribing to our monthly and quarterly magazines, weekly and monthly eNewsletters, and podcast channel.