The robots installed in metalcasting operations are frequently selected for power (lifting molds or heavy finished parts) or reliability (automated grinding) but the recent emergence of collaborative robots — cobots — is demonstrating that robots have feelings too.
Comparable to the emergence of Homo habilis 2.8 million years ago, we have arrived at the point in robot evolution when the brains and the hands are working together. And, that fairly recent breakthrough in robotic functionality has had an after-effect in how manufacturers can apply robots in their operations — and in the software, algorithms, and the next generation of end-arm tooling. The latest wave of end-effector offerings should be particularly relevant to metalcasters, who thanks to these ingenuitive motion and heightened sensitivity, may identify new applications in their operations for this species of automation.
Origin of the Species — The 1980s began with General Motors chairman Roger B. Smith predicting robots would replace humans, as wages rose. A dollar an hour hike would lead to a thousand more robots on the factory floor. His math was off, and by the end of the decade, robot sales were in decline.
"There was stagnation, and robots weren't selling in the way everybody thought they would," recalled Robert Little, CEO of ATI Industrial Automation. It's not that the idea of robots was flawed, as they could handle heavy materials and weld components, but something was missing: versatility. Humans can learn to do pretty much whatever is asked of them, while robots then were fixed, hardwired, inflexible, and due to limited choices for end effectors, severely lacking in dexterity, a feature at which people excel. Often powered by hydraulics, the robots of that generation had a gorilla-like grip. It is why Little, along with fellow engineers Keith Morris and Dwayne Perry founded ATI in 1989.
"We had a mission to develop end effectors that could survive the harsh treatment robots give them," said Little, who started as ATI's COO. "We built ourselves on engineering bullet-proof end effectors, and it worked. The quality went up, and our sales started to rise. It made robots easier to use and increased robot sales in general."
New robot installations in U.S. operations increased steadily in the early 1990s, nearly doubling from 1993 to 1995, when the market first crossed the 10,000-unit threshold. How much of that can be attributed to end effectors is debatable, but what's not is that the robot attachments improved in quality and capability. Grippers, now powered by pneumatics, picked up parts with greater precision.
Adding force-torque sensors led to better feedback, further refining accuracy and repeatability. Interestingly, six-axis force/torque sensors were used as early as 1973 at NASA's Jet Propulsion Laboratory, as JPL researchers noted at the time:
“The drive motor torque required to achieve the link angle/time function is then computed based on a model of the manipulator hardware. The motor voltage/torque characteristic is also modeled, and the computed torque is converted to applied motor voltage. This voltage is computed at a very high rate, applied to D/A converter, and finally to the motor input.”
ATI, now a global leader in six-axis F/T sensors, has advanced the initial concept, culminating in its new Axia80. The high-powered, compact attachment couples to the robot arm and can fit a number of end effectors to increase dexterity. Whereas traditional robots are seen as mechanical and rigid, attaching an Axia80 achieves acute responsiveness that is ideal for high-precision operations. This makes it the perfect accessory for cobots like Universal Robots' UR-3, 5, and 10.
Cobots like the new E-series UR models include F/T sensors already, but the Axia80 sensor is up to five times more accurate, with as much as 20 times the resolution. “It's always good to ensure you have the best force sensor available," Little observed. "The Axia80 detects something as simple as a finger caught underneath a part."
In this regard, the creation has surpassed the creator. Although a human hand may cause a tiny amount of deflection, the Axia80 can sense these minimal variations and signal the robot to react.
Keeping all workers safe is a top priority of any automated cobot system, followed closely by improved performance. F/T sensors are able to improve safety and process flexibility simultaneously. Better sensing makes way for a wider variety of applications, which is an excellent development for those who invest in robotics. As technology advances, there will be more opportunities for robots and humans to work together safely and efficiently.
Unlike the GM chairman’s forecast almost 40 years ago, the jobs these robots take are not the ones people will miss. A worker would be much better suited to managing several machines and talking to current or prospective customers than grinding and polishing a metal part. "Robots always succeed in doing work people don't want to do," Little said.
Parallel (Gripper) Evolution — Three decades of experience support ATI's initiatives. Still, innovation can come from anywhere, at any time, even from a company that has existed for about six months. That's how old the Danish firm OnRobot is: the end-effector and sensor manufacturer fused three startups—from Denmark, Hungary, and the U.S. —to bring machine shops and other automation users three new solutions for tending and assembly. "They go hand-in-hand with the collaborative mindset of easy programming, flexible equipment and being able to use robots in different tasks," said Kristian Hulgard, OnRobots' general manager for the Americas.
Expect the Gecko Gripper to generate a lot of manufacturing buzz, too. Not to be confused with the biomimetic grippers Festo has become known for, this attachment doesn’t immediately resemble a Gecko's foot, though it works the same, using a quantum dynamics phenomenon called the van der waals force. It's complicated, but the short explanation is that s a polarity, and weak attraction, forms between two objects. Like the early force-torque sensors, this functionality was developed by NASA's JPL too, this time as an alternative to Velcro.
"The Gecko gripper works completely differently than anything we've seen on the automation market," Hulgard asserted. Though it was launched officially in North America at IMTS, select customers have been using it on steel, glass, plastic bags and even printed circuit boards (PCBs).
The RG2-FT smart gripper, an evolution of the previous RG-2 and RG-6 servo grippers, employs a six-axis force torque sensor in each "fingertip."
"We can now [be used in] new tasks and start being intelligent," Hulgard said. "We give the robots so much feedback and intelligent sensing that it's opening up new doors."
One of these is in circuit board assembly.
"Imagine, if you need to put a pin in a very tight hole, and use your wrist, not your fingers," Hulgard said. Built-in proximity sensors also zero-in on the part before touching it, to ensure a proper grip automatically. This also allows the wrist to tighten its grasp on electronics manufacturing, which cheaper labor in Asia has wrested away.
"By giving robots the sensing abilities and being able to automate more, that gives us the competitive advantage," Hulgard said. "This keep jobs and keep manufacturing where we want it."
The final new offering, the Polyskin Tactile Gripper, uses sensors on the fingers to offer a less refined, but effective solution for assembly and bin picking. Some features include individual alignment of each finger and bump detection. It's ideal for sensitive or irregular workpieces.
The economical alternative "might not have all features, but you get the major functionality, such as force control and variable grip distance," Hulgard said.
OnRobot also demonstrated the RG-2 and RG-6 servo grippers, which along with the Dual RG2 Dual gripper, have been available for about three years. The Dual Gripper specifically has cut robotic cycle times by 40% and helped a Danish high-mix, low-volume machine shop save time and money.
In that instance, the shop’s operators noticed how inefficient it was for a robot to remove a finished part from the machine and set it down before picking up the next part to work on.
"When we put two grippers on, we could grip and place the object in the machine in the same cycle," Hulgard said. "That cuts out a movement in and out of the machine."
Hulgard stressed that as these humanlike end-effectors are performing as the cobots' "hands," they are augmenting the workers efforts, and allowing for more work cycles to be done.
As cobots' “brains” evolve to a more general intelligence (as anticipated during in the next decade) all this could change; and human workers will have to change, too. Survival of the fittest and all that. But that's the subject of a different argument for another time.
What isn’t debatable now is this: any manufacturer using, or planning to use, the current generation of cobots must be just as adept at knowing what end-effectors are available for their specific task.
"The choice of end effector will start to mean more and more," Hulgard said. "Now, we're in that time when we have to educate the market — and show them these new possibilities."
All of this evolution may be revolutionary in metalcasting, where skills are in short supply and productivity demands are rising. Applications like sand-core placement, or custom grinding and finishing are tedious, but demand focus, speed, and reliability. Now, there are workers perfectly adapted to such jobs.
John Hitch is Senior Technology Writer for IndustryWeek and New Equipment Digest. Contact him at [email protected]