Robotic Metrology: Closer and Closer
- October 8, 2019
Robotic metrology continues to make strides, but challenges remain
When people think of adding robots to metrology, probably the most common application that rings to mind is a probe mounted on the end of a robot arm. But in fact the most common application of robotics in metrology has the robot loading and unloading a CMM or other metrology system.
Starrett demonstrated its own robotic metrology solution when it added robotic loading and unloading to its HVR100-FLIP large FOV (field of view) benchtop video measurement system for a demonstration at IMTS 2018. The HVR100-FLIP has automatic part recognition, so parts can be placed anywhere in the FOV without any fixturing and the system recognizes and inspects the part in a few seconds.
“A lot of the challenge with using robots is parts handling,” says Mark Arenal, general manager of Starrett Kinemetric Engineering Inc. “While there’s a wide variety of grips, a robot has to be able to pick up a part and place it correctly. Once that can be done consistently on the work stage, then it’s pretty smooth from there.”
For a loading/unloading application, the integration of robotics and metrology is also a pretty straightforward matter, Arenal says. “Once the machine program finishes running there’s a signal sent out to the robot to put the part in the right bin. To set this up properly, there’s a bit of upfront work required by an engineering or quality person, but it’s relatively simple.”
One of the main advantages of this scenario is that a single operator can setup the metrology and loading/unloading routine and once that has been done, oversee several such systems at a time.
“Normally the first step in adding automation to a CNC machine is to have it loaded and unloaded by a robot,” says Jamie King, regional manager, Canada for Blum-Novotest. “Oftentimes the next step is to add metrology to ensure good parts are coming off the machine. To close the loop you want the information from that gauge to be fed back to the machine so that you keep adjusting to make the most accurate parts possible – and also to stop when it’s no longer capable of producing a part in tolerance.”
That second step—the addition of robotics to the actual measurement process—is where robotics truly has the potential to transform metrology. While it has yet to become commonplace, the potential is well known. Blum-Novotest showed what could be done at IMTS when it demonstrated its BG60 bore gauge measuring parts while mounted on a Fanuc robot arm.
The demonstration highlighted the fact that metrology devices robotically mounted will need to be self-calibrating, as robot path finding isn’t yet precise to the degree required to measure a part independently. The BG60 is one example of a contact measurement system that can be adapted relatively easily for use with a robotic system. Most systems for robotic measurement use laser and other visual scanning methods.
Ravi Prasad, sales and application specialist for Mitutoyo Canada, recalls one customer that used Mitutoyo’s Surftest SJ-410 surface roughness tester on an automotive component with some 50 surfaces that had to have surface finish detected.
“A person would have to move and manipulate the part for every one of those measurement points,” Prasad says. Mitutoyo had a fixture developed to flip the part halfway through the measurement cycle and mounted the SJ-410 to an ABB robotic arm.
“Literally as far as the operator was concerned, it was load the part onto the fixture and go,” Prasad says. “They had to do something that was quicker than a human.”
In cases where the decision isn’t forced on the customer, anyone contemplating the addition of robotics to their metrology process has to take other factors into account.
“You have to consider the repeatability,” Prasad says. “You’re looking at a robot presenting the metrology equipment to the part at the exact same spot every time; no delays, no coffee breaks.”
Warren Reynolds is general manager at I-Cubed, based in Stoney Creek, Ont. I-Cubed specializes in developing industrial robotics solutions. Reynolds says initial expense shouldn’t be an obstacle, and he believes the big plus for robotics—for any application—is that the systems can be repurposed easily once set up.
“People come in saying they need a robot and what’s the price tag. When we say $150,000 they say ‘I thought the robot was $80,000.’ That might be the price for the robot, but by the time you build everything you need around it, such as end-arm tooling, have it safety certified, have it communicate with everything, remote access, etc., that’s where the price tag comes in. But once it’s in, it’s super flexible.”
That said, robots aren’t yet suited for all kinds of metrology. One problem is that the typical path accuracy of a robot isn’t yet as high as it needs to be to be applied in a metrology setting.
“If you want to do a circle, you can take points on the circle or you can just run around the circle and do a constant scan,” Reynolds says. “The problem with that constant scan is that the path accuracy margin of the robot could be larger than the error you’re looking for in the hole. So your gauge repeatability and reproducibility is shot.”
That means that so far, robotic metrology is confined to more or less portable optical measurement systems that have been developed to cope with the margin of error inherent when devices are being held by a human operator.
“We played with adding a Creaform MetraScan laser scanner to a robotic arm,” Reynolds says. “Now Creaform has the unit for sale. They use an external base scanner to locate the head in space and take the scan data from there. That takes out all of the robot’s inaccuracies.”
The Creaform system is based on the MetraSCAN 3D-R optical 3D laser scanner mounted on a robot arm. The system also has a C-Track optical tracker which, mounted nearby, automatically detects part alignment and movement. The R-Series autocalibration kit automatically calibrates the C-Track to the laser scanner so that the whole system can compensate for inaccuracies.
Reynolds is careful to dispel the common misconception that robotic metrology is difficult to master. Path accuracy may still be a stumbling block, but beyond that, it’s not as hard as people think.
“You learn by doing it. Among my operators I have one guy who was originally trained to do HVAC. Another did home renovations before coming here, and another was in small engine repair. They’re young and they pick this up quickly. There’s only one operator here who was actually trained as a robot programmer.”
Reynolds says he’s put in robotic systems for customers who say at first that they don’t like it and aren’t getting what they want from it, and about a month later have become comfortable with the technology. With the development of robotic systems that can be programmed like conventional CNCs, the entry barrier has declined even more.
“A normal three axis CNC gives you X, Y and Z. A robot has six axes and an angle on the tool at all the time, so there are conversions to be done, and they can be done. There’s now software that can do it.”
Reynolds is optimistic that great strides will be made in applying robotics to metrology, but he cautions that it probably won’t happen as soon as people might think.
“For now a lot of what you’re going to see is on the machine tending side; that’s where we’ll see the biggest jumps first. After that it will be when somebody comes up with better controls and checks and balances on position to allow for better inspection. But everyone’s making strides all the time.” SMT