Move over machine tools here come robots

Feb. 17, 2006
Smooth translations catapult robots into five-axis machining cells.

Robots do 5-axis machining with CAM-software G-code translated into robot language.

More shops are using robots instead of machine tools for secondary operations.

Shops no longer "teach" robots to move from part to part.

There is a classic failure to communicate in the machine shop. Robots run on brand-specific languages all their own, while conventional CNC machine tools read G-code generated through CAM software. But what if a robot's control was programmed to read G-code directly from a CAD/CAM program?

That robot could gain the functionalities of a CNC machine tool, including full 5-axis contour surface-machining capabilities, and it could be programmed in minutes to work on any 3- or 5-axis workpiece. Current technology typically requires hours, or even days of programming before a robot is ready to work. In addition, machining cells that included such robots could provide a low-cost alternative to machining centers and routers for secondary part operations and could eliminate the work-envelope constraints of those machines when dealing with large components.

Programming Plus Inc. ( says its Robotic Machining Cell turns this "what if" scenario into a reality. The robotic machining cell hinges on software developed by the New Berlin, Wisc., company that smoothly translates CAM-generated G-code that is ready for a 5-axis machine tool into usable code for a 6-axis robot. No other company in the world has accomplished that, says Tim Brooks, robotic sales engineer at PPI. Programming Plus specializes in CAD/CAM, DNC and shop floor automation.

Programming Plus uses Delcam ( Power-Mill software to generate G-code for Kuka Robotics ( robots in its robotic machining cells (RMCs). Since the cells function like machine tools, operators can turn the robot's spindle on and off, change rpms, and override speeds and feeds. Until now, such functions could not be done, says Brooks, because they were not within a robot's control.

Shops typically use conventional robots equipped with spindles for trimming, cutting and deburring around part profiles. But these shops must "teach" robots by moving them to multiple points around parts, and the more complex the part, the longer teaching takes. And teaching robots to go from one part to a different one requires even more teaching. Robots in a robotic machining cell move from one part to the next without having to be taught, and do so with less than 30 minutes of preparation time, no matter what the level of part complexity.

If an operator is unsatisfied with a robot's cutting path in a robotic machining cell, he can change the cutting path in the CAM system or he can manually teach the change and insert it into the toolpath, resave the program, and run it. Also, since robotic machining cells use Delcam's PowerMill software, shops can conduct robot and part-machining simulations to check for possible collisions and obstructions, and do gouge checking prior to running a program. In addition, Brooks says his company's software soon will let shops monitor robots, via cameras, from remote locations over the Internet.

Programming Plus' software offers shops another benefit: resolving problems of singularity. Singularity occurs when all of a robot's axes approach being in line with zero. They increase their speed to keep up, they get to a point where they are moving too fast and, in a way, they lock up and stop. The condition is similar to constant surface footage on a lathe. The problem is that shops can not predict where singularity will happen.

Fortunately, as the robotic machining cell software transfers G-code to a robot, it evaluates the program for physical configurations that might cause singularity. If any are found, the software modifies the program to eliminate them.

Most shops do not associate 5-axis contour surface machining with robots primarily because it would take a great deal of time to teach a robot all the necessary data points. Instead, shops use CNC machine tools. If the parts are large, they are split into multiple pieces that fit on the machines and reassembled after cutting. Alternatively, shops will remove machine panels and doors to mount the portion of the workpiece that needs cutting on the machine, while the rest of the part hangs out or over the sides.

Shops with robotic machining cells can work on whole parts in single setups because the cells do not have exactly defined work areas as machining centers do. Parts can rest on a shop floor or they can be clamped to workholding tables and, as long as the setup is within the robot's reach, they can be machined. In some instances, the robot's reach can extend to 25 ft.

The current ideal application for a robotic machining cell is one that involves large, complex parts to be machined on a 4- or 5-axis machining center that are made of soft materials such as plastics, fiberglass, carbon-fiber composites and the materials used in prototyping. However, Brooks expects to see his company's robotic machining cells being used to machine soft metals by the end of this year.

When it comes to machining metals, there are two factors that hold back most robots: accuracy and torque. Today's robots, on average, are accurate to about 0.004 in. Although most metal parts require precision machining, there are applications in which robot accuracy is acceptable, such as rough machining parts to prepare them for finish cutting on CNC machine tools.

The increased torque that robots require to machine soft metals can be had with more-powerful spindles, but robots have to be strong and rigid enough to handle those spindles, Brooks notes.

Programming Plus uses Kuka robots in its robotic machining cells, and Joe Campbell, director of strategic alliances at Kuka, says several technologies support robot machining. For instance, companies such as Kuka now build robots more mechanically stiff and rigid than they ever have been.

Tighter tolerances in gear trains reduce robot backlash, while finite element analysis and computer simulation tools have helped to remove flex from robot castings. Robots are more intelligent and display tight motion control, and builders can map their inaccuracies better than they were able to in the past.

"We now have a robot platform worthy of machining operations," says Campbell, "It may not sound like much, but considering the mass of a machine tool as compared with that of a robot, today's robots are quite rigid and strong."

Kuka offers a handful of robots for machining, including its KR60-HA, KR210 and KR240 models. The KR60s are currently Kuka's most accurate robots because they feature matched gear sets and other assembly-process techniques that calibrate them for high accuracy. Campbell explains that there is a tradeoff between the material a robot removes and the level of precision it can hold when machining hard materials. A shop that cleans mold marks and flash from die-cast motorcycle wheels, for example, might use a CNC machine tool, but that operation is perfect for robots because it requires precision and minor amounts of material removal. On the other hand, he says that cutting keyways in hard stock is not an ideal application for a robot. " Robots could not do most work that is done on machine tools. But, there are a lot of shops using CNC machines for secondary processes that robots are perfectly capable of doing," says Campbell.

KR210 and KR240 Kuka robot models handle big payloads. And, while milling spindles are not all that heavy, the robots pack ample motor power to maintain position when traveling along cutting paths. "Machine tool builders talk about spindle horsepower. We in the robot business focus on payload capabilities," Campbell says.

The Kuka robots used by Programming Plus easily carry 30- and 40-horsepower spindles. Kuka's strongest robot hefts payloads weighing up to 1,100 lbs. within its 25-ft. reach, and the company is working on even bigger models that Campbell says will open up more possibilities in material-removal applications. Other Kuka prototype models currently in testing feature low travel speeds but extremely high torque and rigidity to further enhance robotic material removal.

In machining applications, robots often use force and torque sensors that allow them to "feel" incongruence or imperfections in surfaces and remove a minimal amount of material to correct them. For example, to blend two surfaces, a robot automatically adjusts the applied torque and force to help it to avoid damaging the surface or removing too much material from the part. Force and torque sensors integrate into a robot's wrist, and the robot controller oversees their operation. Those sensors and the controller combine to allow users to specify the direction in which the robot will apply force, Campbell says.

A typical robotic machining cell consists of a robot with a 5- to 10-horsepower spindle, a covered toolchanger rack that holds 10 tools, a device for determining a loaded tool's length and a laser system that locates workpieces for easy part-to-part setups. Spindle size depends on the application.

A Programming Plus-system robot can change tools automatically. Campbell says toolchanging on the robot side is common, and the company's standard millhead toolchanger is straightforward and simple. This capability makes cell robots self-loading. They remove their milling tool, load a part into the cell, re-attach the tool and cut the part.

"The motivation we hear from customers interested in a RMC is throughput. Most CNC machine tools, unless automated, have low utilization rates, meaning their mill heads are not turning all the time because of setup time and other non-cutting activities. Our robotic machining cells, on the other hand, easily accommodate multiple parts within their work envelopes, allowing one robot to move from part to part for significant throughput at less capital investment," Campbell says.

Latest from Machining / Cutting