Robotics adoption: The SMB guide to industrial automation
Robots aren't the exclusive purview of major enterprises anymore. Small and medium-sized businesses are finding ways to save money and streamline operations by automating. This guide will show you how.
Since the first industrial robot, Unimate, was installed at a GM plant in the 1950s, industrial automation has been associated with big businesses running huge operations that involve massive production lines. The size, shape, and dynamics of industrial robots long reflected this reality. Industrial robots were big, noisy, massively powerful, and required their own cages to keep them well away from human workforces.
A radical transformation in industrial automation has occurred over the past decade, and it's as much a technological shift as a consequence of the changing economy, one increasingly reliant on small runs, fast shipping, and nimble operations. A new generation of robots reflects these changes. They are quickly deployable, task-agnostic, smaller than their clunky forebears, and can work alongside humans outside of cages.
To be sure, huge companies like Amazon, which acquired logistics robot maker Kiva for a staggering $775 million in 2012, have driven the shift in industrial automation. But the shift toward nimble operations has resulted in the recent development of countless tabletop and mobile robotic platforms that may be surprisingly appealing to a new kind of customer: Small and medium-sized businesses.
The indicators are clear. The fastest growing segment of industrial automation is collaborative robots, so named because they can be deployed to work alongside humans outside of cages. One major supplier, Yaskawa Motoman, estimates that 95 percent of its robotics customers have five robots or less. Compare that to the old days of industrial automation when dozens or hundreds of task-specific robots were required to produce a finished product and you begin to see the magnitude of the shift.
This guide will serve as an introduction to industrial automation for customers who may never have considered automating part of their operations. The opportunities are huge for those savvy enough to stay ahead of the curve.
Historically, one of the biggest barriers to entry to industrial automation has been the massive costs and time constraints associated with setup. It used to take weeks or months to set up and dial in an automated production line. The task was performed by a team of robotics experts adept in programming languages and advanced engineering, subjects unfamiliar to most.
Perhaps the most significant change to come to industrial automation in the last few years is that robots have become readily deployable into lines without dramatically slowing down or reconfiguring operations. And thanks to advances in how robots learn, many systems are designed to be set up by non-experts.
The term of art is programming-by-demonstration (PbD), which is exactly what it sounds like. Humans physically move the robots through actions, which the robot can then repeat. According to Universal Robots, the market leader in collaborative robotic arms:
Patented technology lets operators with no programming experience quickly set up and operate our cobots with intuitive, 3D visualization. Simply move the robot arm to desired waypoints or touch the arrow keys on the easy-to-use touchscreen tablet.
Universal Robots has revolutionized cobot set-up, reducing typical robotic deployment measured in weeks to a matter of hours. The average set-up time reported by our customers is only half a day. The out-of-box experience for an untrained operator to unpack the cobot, mount it, and program the first simple task is typically less than an hour.
Usually, PbD isn't seamless, often requiring adjustment via an API. Universal Robots ships its units with teaching pendants, portable programming consoles aimed at novices. Consultants and a rapidly growing field service sector targeting small and mid-sized customers are cropping up, as well, and many suppliers offer help getting their robots online, easing the transition to automation.
The most popular collaborative robot on the market is UR5 by Universal Robots, so it's a good starting point when talking about price. UR5 is a robotic platform with six degrees of freedom and a 5kg payload. It can be fitted with multiple end effectors (the part that actually interacts with the world by grasping or manipulating objects) and is suitable for a variety of tasks like components manufacturing, pick-and-place, and benchwork in labs and light fabrication shops.
UR5 costs about $35,000, including most components and the teaching pendant. That puts it middle of the pack in terms of collaborative robots. Oftentimes, additional grippers or end tools will be needed for specific applications, driving up costs.
With the leaders in collaborative robot platforms solidifying, it's become a formative time for the end effector market, with companies rapidly joining forces and innovating to gain market share. There was recently a three-way merger in the "robot hands" market as Perception Robotics (US), OptoForce (Hungary), and On Robot (Denmark) came together to form OnRobot. Soft Robotics, a company using materials science in favor of complex engineering, makes soft grippers that conform to objects and are suitable to a wider range of tasks than their hard-skinned counterparts.
This one carries an asterisk. Collaborative robots are billed as safe, and by and large that's true. But a common saying in the industry is that a robot is only safe until you put a knife in its hand.
In other words, no tool is either safe or dangerous inherently. It's a matter of how it's used. There's currently a groundswell effort in one corner of the industry to help establish stricter compliance protocols to ensure safety.
"I honestly hope 2019 is the year that robot safety becomes a top priority," Melonee Wise, CEO of Fetch Robotics, which makes mobile collaborative robots, told me recently. "The potential mistakes of one robotics company could set our industry back for years."
In the US, workplace automation safety guidelines issue from OSHA, but those guidelines haven't caught up to the rapid evolution of the technology and it's still a bit of a Wild West environment when it comes to uncaged robots.
All that said, the safety record since robots have come out of cages to start working alongside humans has been strong. A combination of force sensors, which cause robots to stop when they bump into things, advanced machine vision, and other sensing innovations have resulted in platforms that are reasonably safe if deployed responsibly in an environment where employees understand the risks.
Does your company warehouse and ship components? Manufacture dongles? Perform scientific testing? Fabricate metal? There's a robot for that. In fact, it's possible all of those operations could be served by the same robot.
The trend in industrial automation is task-agnostic platforms that can be customized and configured for an infinite number of deployment types. Companies like Fetch Robotics, which makes autonomous mobile robots for materials handling and picking & placing, can move materials in a warehouse just as easily as they can stock inventory in a logistics operation. When designing a lightweight pneumatic collaborative robot, Festo intentionally modeled it after a human arm, a nod to the range of tasks it was designed to perform.
The widest integration of collaborative robots has happened in industries like components handling, but as you see in the use cases and case studies outlined below, SMBs of all stripes are finding innovative ways to automate with collaborative robots.
What tasks can be automated?
Repeatable tasks in predictable environments. That could mean picking diverse products off a line and placing them in a bin. It could also mean operating existing machinery, such as CNC machines or welders. At the other end of the spectrum, it could mean holding a beaker and pipetting chemicals.
Automation is so versatile that it's difficult to home in on just a few sectors where SMBs are benefiting from automation. This list isn't exhaustive, but it will give you an idea of the diversity of applications for industrial automation.
Of all SMBs, fabricators are probably most familiar with automation thanks to CNC machines. But industrial 3D printers, such as HP's Metal Jet 3D, which can print metal components, collaborative robots capable of tending CNC machines, and materials handling autonomous mobile robots, such as those from Fetch Robotics, are expanding the automation opportunities in the sector.
Of all the sectors that are now adopting industrial automation and robots, agriculture has to be the longest coming. That's because plants are fragile and tasks that are simple for humans, such as picking a strawberry, present incredible machine vision and path planning problems for robots.
But new solutions are now available. Soft Robotics, which makes flexible end effectors, has used materials science to help solve a persistent issue in industrial automation: Hard robot hands tend to bruise delicate objects like produce. (The CEO of Soft Robotics, Carl Vause, recently welled with pride while telling me how his company's products finally managed to automate part of the Peeps line; no other company could produce an end effector capable of repeatedly picking up the marshmallow treats without ruining them.)
Automated tractors and water monitoring solutions are now becoming commonplace in agriculture. The legal marijuana industry is the latest to explore automation. A company called bloom has created an automated picking solution that won't damage the crop.
Construction sites offer a nice compromise between the structured world of factories and the unstructured world of roads and highways. The sector has recovered nicely since the collapse a decade ago, and a spate of new and legacy companies has taken dead aim at the $1.2 trillion industry.
Caterpillar, for example, recently invested in Sarcos, a robotics company with a line of exoskeletons that augment human strength.
Companies like Ekso Bionics, which started out catering to the medical and rehabilitation markets, has tacked toward heavy industry with a new line of products that allow workers to hoist tools above their heads for hours at a time without injuring themselves.
And there are now several companies that use drones to track progress at construction sites in real time.
Silicon Valley players like Built Robotics are figuring out how to turn dumb machines like bulldozers into autonomous, data-gathering systems to streamline site operations.
A list of common causes for error in pipetting include failure to wet the pipette tip, failure to take the temperature of a substance into account, failure to wipe pipette tips after working, pipetting at an angle ... it goes on.
Introducing error at the measuring stage is particularly insidious, as those mistakes are difficult to track down. Following a trend we're seeing with cheap table-top 3D printers, which are fast bringing advanced manufacturing to the masses, a company called Opentrons recently announced a new version of its small pipetting robot, which costs about $4000, far less than other automated lab solutions.
Desktop collaborative robots are also helping automate laboratory tasks, serving as a helping hand on a bench.
Automating professional services: TLAC Toronto Printing & Publishing
There's a perception that automation and old world craftsmanship are at odds with one another. But in an era of flexible, easily deployable automation that's just not the case.
I was reminded of this on a recent call with Miraz Manji, founder of TLAC Toronto Printing & Publishing, a 2D and 3D design, print, and publishing shop in Toronto, Canada, that creates books for self-published authors. Manji's team uses a mix of automation, old fashioned design talent, and close attention to craft to produce small and medium-sized runs of physical books of exceptional quality -- a niche market until automation increased the number of artisanal books a small shop could produce.
"The book as a product has minor details of creativity that hold high value to the author and reader," said Manji. "By merging custom-manufacturing with robotics, we're keeping those emotional and artistic aspects, such as the coloring of a photo or the kind of paper used. We need humans to do what they do at the highest level, which is design and inspection."
Publishing provides an interesting historical case study precisely because it was an industry once occupied by master printers and bookbinders who were replaced en mass with the arrival of industrial-scale publishing. The old jobs were all but eliminated.
But then the publishing industry changed. "The industry went through a major shift and a lot of people self-published," explains Manji. "The old robotic process of producing ten thousand or fifty thousand doesn't work. Now we're down to 250 books. The reaction to (and, reciprocally, a driver of) all this was a proliferation of print-on-demand services. "There were a lot of rookie mistakes. We saw that ten years ago just in creating a book. No one really knew how much involvement a customer should have in the actual prototyping process. How do you maintain those boundaries?"
Quality was all over the map, also, as anyone who's read a friend's self-published book knows. Fluorescent white pages between two pieces of card stock does not a pleasurable reading experience make. "A lot of companies went out of business," Manji said. "We survived. We did it by focusing on our formula, our workflow."
TLAC uses a robotic book binder that stitches, clamps, and glues, emulating a hand. The system is faster and more flexible than the famous Espresso book machines that still churn away in a few book stores across the nation and produce single books to order.
The secret sauce at TLAC, though, is the way they've streamlined their prototyping workflow to take advantage of software and robotic automation while preserving their allegiance to craft. "We have a background in process automation," Manji said.
TLAC's clients are matched with designers and editors, who create design files that are cleaned up using an automated process. After that they're touched up by a human during inspection. TLAC uses its book binding robot in conjunction with human labor to produce a prototype. The process isn't fully automated, and a worker can intervene and make corrections on the fly. "We throw out the first fifteen books. They're just not there for what we call 'printing that matters.'"
When a prototype is deemed acceptable, larger-scale printing begins. The pages are cut using a programmable cutter, given another quality check, and bound by the robot. TLAC developed a proprietary sensor glove that a worker can wear to count finished books -- a novel solution it may license. The client receives automatic updates as their book progresses through TLAC's 11-stage process. The amazing thing about this isn't that the individual pieces of technology exist -- they have for a long time -- but that TLAC is a small business with five employees. By taking advantage of the best of human craftsmanship and the most accessible manufacturing and workflow automation tools, this little startup is helping redefine its market while pointing to a new way for savvy small manufacturers to deliver both value and quality.
In the US, domestic production of eyewear pretty much dried up in the 1980s as foreign concerns acquired American companies and off-shored manufacturing to save money. So when architect Brian Vallario and finance guy Gerard Masci decided to open an eyewear company built on products Made in the USA, there was all the reason in the world to be skeptical of the effort.
The Brooklyn-based company they started is called Lowercase. Vallario had done some research on robots in graduate school and had enough passing familiarity with automation to know the technology was changing and could offer tremendous benefits to a small manufacturer. The team sought out Axis, a company that helps small businesses integrate industrial automation.
"It was a complete learning process but it went fairly smoothly," he recently told Robotiq in a case study titled Never Too Small for Cobots. "With the help of Axis New Jersey, we were put in touch with an integrator but the majority of the integration was done on our own."
Axis recommended using an arm from Universal Robots with a gripper from Robotiq. A local integrator was brought in to create a custom stand where the robot stands. The robot was configured to interface with a CNC machine that cuts out glasses frames from solid sheets. The entire integration took five days.
"The cost to automate was extremely high," Masci told Robotiq, "but we were always able to get our heads around that because of the value that the robot would add to the process. The main problem was that the integration cost of most robotic solutions was 3-4 times the cost of the solution itself. When we worked out a solution using the UR5 and the Robotiq gripper, the integration cost was a fraction of the actual purchase price. Then it made it tenable for a small company of two people to integrate a robot into our solution."
The final configuration starts when Large sheets of material are cut down to smaller more manageable size. The Universal Robot arm tends the CNC machine, loading the sheets to first position, then to second position. Each Lowercase run averages about 500 units with several style iterations within each run. The robot loads the cut frames into a bucket for finishing, completing the automated process.
"We have 17 different styles now," said Vallario. "We really needed something that was easy to program and that we could manipulate on the fly. If I make a change on a design, I have to be able to apply the changes quickly. And since we are competing against bigger, high-end eyewear manufacturers, we have to deliver the highest quality even if we do quite smaller productions."
Old school fabrication, new automation technology: Goldbrecht USA
Manufacturing is a way of life in Los Angeles, a city that accounts for 523,100 manufacturing jobs, more than any other metro area in the country. Much of Los Angeles' industrial base comprises small- and medium-size businesses, which makes it an important bellwether for the health of the nation's light manufacturing sector.
I reached out to James Tschortner, CEO of Goldbrecht USA, a manufacturer of door and window solutions in the Los Angeles area. "The main issue is the lack of educated manpower in the market," he said. "When we have interviews, we have people coming in, and we're extremely surprised by what they don't know."
It's understandably vexing to weed through unqualified applicants. But pull the lens back a bit and the lack of qualified manpower should come as no surprise. The twin birds of economic prey, off-shoring and automation, went after low-skilled manufacturing jobs first. Middle- and high-skilled jobs were more resilient until the recession took a big bite; all told, we lost about 6 million manufacturing jobs in the US between 1990 and 2012, according to a report prepared last year by the Los Angeles County Economic Development Corporation.
Over the same period, however, national manufacturing output grew by 46.7 percent. In other words, productivity skyrocketed. In durable goods, labor productivity rose 142 percent between 1990 and 2011, a clear signal that manufacturers have been investing in technology over labor.
That's been Tschortner's approach. With targeted investments in technology, his company has been able to keep pace with demand, despite not being able to find qualified workers. But now that construction is once again booming -- translation: lots of orders for doors and windows -- he needs to ramp up operations quickly. In a tight labor market, Goldbrecht USA has turned to automated CNC machines and tabletop robots. That's helped the company overcome labor obstacles that in the past would be strict inhibitors of growth.