Turning Out the Lights on the Factory Floor

Though true lights-out production is still rare, more processes are running with limited human interaction. Variability, complexity, dexterity and danger shape decisions.

Will the factory of the future be completely automated, with only the proverbial man and a dog—the man to feed the dog and the dog to keep the man from touching anything? It’s hard to tell at this stage in technology development, but it’s absolutely true that more and more production processes, if not entire factories, are running with limited human interaction.

The evolution to more automated processes has been years in the making. Although many of the technologies needed to automate manufacturing work behind the scenes—such as programmable logic controllers (PLCs), software and vision systems—robots have become the most visible symbols of lights-out manufacturing.

“Robots are uniquely suited to any manufacturing task that is dirty, dangerous or dull,” says Douglas Peterson, general manager for the Americas at collaborative robots maker Universal Robots. “If the process doesn’t require human dexterity, mental agility or problem-solving skills, then it can be done by a robot.”

One of the most widely known examples of a completely lights-out factory is a Fanuc plant in Japan where robots make new robots without human intervention. It’s easy to see why a maker of robots would want to demonstrate the lights-out concept, but many manufacturers are exploring the possibilities of using this approach for at least some of their production processes.

Although true lights-out factories are still rare, the use of collaborative robots is growing rapidly, particularly in third-shift operations. “You have to specifically design a factory floor for lights-out operation, which requires a significant capital expenditure,” Peterson says. “But collaborative robots can be easily integrated into the existing factory layout because they’re designed to work safely alongside humans. They can also function without human supervision.”

There are a number of reasons why companies might consider lights-out processes. For some, the goal is better product quality and throughput rates. Others want to reduce workplace injuries, particularly in hazardous environments like paint shops, where high temperatures and noxious gases can pose human dangers. Another factor is the difficulty in finding workers, with an estimated 2 million manufacturing jobs going unfilled today.

Attracting skilled employees is particularly difficult for the small and medium-sized companies that make up the bulk of the country’s manufacturing base. According to the National Association of Manufacturers, of the 252,000 manufacturing companies in the U.S., only 3,700 had more than 500 workers. The vast majority of these companies employ fewer than 20 people.

Though machines cannot replace human workers in many production processes, even the smallest companies are finding collaborative robots (co-bots) an easy-to-deploy solution for this worker shortage. Ironically, co-bots may also have a role to play in reindustrializing America and making it possible to create new manufacturing jobs.

Robots as partners
Consider the experience at Creating Revolutions, a startup making a customer service paging system for the restaurant industry. Customers activate the device using their smartphones, cueing busy waitstaff when service is needed. The system also records data points like response times, helping restaurants better manage their employee resources to improve customer satisfaction and turnover.

A collaborative robot at Creating Revolutions works side by side with humans during the day and in lights-out mode at night.​

Creating Revolutions was experiencing double-digit product rejection rates because of faulty assembly. It was also unable to find skilled workers. Seeking an alternative solution for these production problems, founder Einar Rosenberg turned to Hirebotics, another startup that offered to rent a collaborative robot by the hour. Renting from Hirebotics eliminated the capital expense barriers for new technology faced by nearly every startup. Using the robot also reduced product rejection rates to nearly zero. Perhaps the most surprising result, the company was able to add manufacturing jobs.

“We wouldn’t have been able to afford to make our products in the U.S. without the robot,” Rosenberg explains. “We’ve even added jobs to keep up with the increased production the robot generates.”

The company can also make changes to its manufacturing processes and products on the fly, a common need at startups that would have been impossible to achieve with offshore manufacturing.

In addition to functioning side by side with human workers during the day, the robot continues to perform for hours after workers go home at night, preparing parts for assembly the next day. Creating Revolutions expects to add three more co-bots in 2017 to keep up with demand.

The machines-as-a-service business model used by Hirebotics is becoming more common for machine builders and automation suppliers in the face of ongoing customer resistance to capital expenditures. Hirebotics purchases its collaborative robots from Universal Robots and uses the cloud to program and monitor their performance for customers, as well as to provide ongoing technical support. Performance data is streamed from the robot as events occur and can be monitored by Hirebotics with a mobile app. Web cameras show what’s happening on the production line, allowing customers as well as Hirebotics to monitor activities at all times.

“We see our job as helping manufacturers in the U.S. succeed,” says Matt Bush, Hirebotics COO and co-founder. “Most companies are years away from lights-out manufacturing because it’s so capital-intensive. But by renting robots by the hour and only paying for the hours they actually work, companies of all sizes can afford to automate more of their processes.”

Bush adds, “While robots work more slowly than human workers, they also work more consistently, without disruptions or distractions, and for more hours, so 10-20 percent increases in productivity are common. Robots can take over the boring, repetitive tasks, freeing human workers for more interesting jobs. People who use them are beginning to see robots as their partners, not their competitors.”

Keeping things moving
Although robots have been tending CNC machines, loading and unloading parts, without human intervention for years, customers today want greater flexibility and mobility. WhenWhippany Actuation Systems, a maker of specialized electromechanical actuation systems for the aerospace and defense industries, needed to increase production quickly, it looked for an alternative to sinking a large expenditure into a new CNC machine. The solution was a robot from Universal Robots that could tend its CNC machine overnight. It’s able to handle parts of different sizes, using customized adaptive grippers, and communicate with both grippers and the CNC machine.

Customized grippers on this Universal Robots unit let Whippany Actuation Systems run two unattended shifts to meet increased production demand.​

“We wanted a solution that could be implemented and programmed easily and didn’t require traditional guarding and safety,” explains Phil DeMauro, Whippany’s manager of manufacturing engineering. “The ability to move the robot from one cell to another, just by popping a couple of locating pins in the floor to get up and running, was also a huge consideration.”

With Whippany’s CNC machines now running two additional, unattended shifts, the company has met its goal of increased production capacity and DeMauro expects to achieve ROI on the robot in less than a year. “Overall, it helps the business from a cost, productivity and capacity standpoint, and we expect it to improve our margins.”

Another mobile solution—one that can actually move anywhere on the factory floor unattended—is the MiR100 mobile autonomous robot from Mobile Industrial Robots, a Danish company that entered the U.S. market early in 2016.

“A mobile platform is the single most important recent development in robotics for material handling,” says Ed Mullen, vice president of sales for North America at Mobile Industrial Robots. “Traditional automated guided vehicles (AGVs) require fixed tracks or sensors installed in a factory floor, making them expensive and inflexible, as well as unable to avoid obstacles in their path.”

Self-driven mobile robots, such as this one from Mobile Industrial Robots, can avoid obstacles and change paths as they carry goods through a factory.​

Upon installation, a mobile robot is taught the layout of a factory and the paths around it, so it always has a context within which to work. Once programmed, the robot can find its waypoints without human interaction. The operator interface is handled through a wireless web interface. Additional modules can be installed on top of the robot to meet various application needs, such as a 24-inch conveyor to move goods without human assistance from one production station to the next. Other module options for lights-out operation include a lift, a towing hook or even a mounted collaborative robot arm.

Elos Medtech, a Danish maker of medical devices and components, uses its MiR robot to move materials around the factory. Before the robot, staff used to walk an average of 7.5 km a day, pushing goods on a cart between departments. That added up to 1,650 km and many hours every year. Now workers are assured a faster, steadier flow of materials throughout the production process, helping eliminate downtime and latency.

Where does lights out work?
There are many processes within a factory that fit the criteria for a lights-out approach as long as the system is set up properly, according to Will Aja, vice president of customer operations at Panacea Technologies, a member of the Control System Integrators Association (CSIA).

“We’ve done a few projects in the pharmaceutical industry where eliminating human errors is a big concern,” Aja says. “Fully automated storage and warehouse facilities are becoming common for distribution companies like Fed-Ex or UPS. Factories that make products like flour, where there’s a great risk to humans from explosions, are also using increased automation. In addition, we’re seeing hybrid approaches for inspection and quality control that are directed by software but where a human has to take an action.”

Although cost is a major barrier to lights-out manufacturing, Aja sees machines performing an increasing number of tasks in factories around the world. “No matter in what country a product is being made, it’s hard to make things cost-effectively with a lot of workers,” he says. “How far you can go is heavily process-driven. It varies by industry and there’s a range of decisions that have to be made.”

Companies making commodity products might not have the cost structure to support it, Aja adds. “If your brand is based on a level of quality that sets it apart, you may not want to automate. On the other hand, if you’re using expensive raw materials, then the cost of throwing away products ruined by human errors is a big consideration.”

Product variability and environmental conditions are two other critical factors to consider. “Automation often breaks down when you’re making bakery goods or confections, such as when buns aren’t separated enough,” says Mark Noschang, robotics applications engineering lead for Omron. “Produce handling is another difficult area, since no two apples are ever the same. Applications like these often require some manual intervention.”

Packaging can also be tricky to automate. “Plastic clamshells for lettuce or spinach are a particular challenge, since leaves often stick out and prevent the package from closing properly,” Noschang says. “Cold, moist environments that can warp cardboard boxes are likely to need human supervision. It’s difficult for a robot to deal with variations. Even if it can sense a problem, a robot often isn’t capable of correcting it.”

Noschang predicts that industries like solar, automotive, medical devices and electronics will soon automate large portions of their manufacturing processes. “Another trend is the development of autonomous intelligent vehicles (AIVs) for material handling, with built-in safety and navigation features. These will increasingly replace traditional AGVs that require tracks, wires, sensors and magnets.”

As technology changes what’s possible, it will also change business economics, predicts Beth Parkinson, market development director for Connected Enterprise at Rockwell Automation. “Many processes lend themselves to the lights-out approach, such as those involving high temperatures, toxic gases or high payloads, and applications like furnace and paint line management or carbon fiber cutting, which pose potential harm for humans.”

Parkinson cautions, however, that pushing lights out too soon can actually cost more than what a company is paying for labor. “It will be important to prioritize areas that pose dangers to humans, where new workers don’t have the needed skills or don’t want to do repetitive tasks,” she says.

Parkinson adds, “With the increased connectivity of the Industrial Internet of Things, manufacturers will find they need to attract a new kind of worker—one more interested in improving manufacturing technology or managing those processes than in doing manual labor.”

What’s next for robot tech?
As robot technology advances, so will its increased use on the factory floor. “By the year 2020, an expected 150,000 robots will be sold every year,” says Jim Lawton, chief product and marketing officer for Rethink Robotics, one of the early developers of collaborative robots.

“Think of today’s robots as PCs with arms,” Lawton says. “Our goal was to make them more useful and easier to use, as well as more robust and reliable. Software has become more important than hardware, allowing robots to handle a wider variety of tasks. By leveraging technologies that weren’t available 10 years ago, we’re turning robots into teachable machines that respond more like a human.”

Lawton adds, “People are good at dealing with variability, but not as good at looking at large amounts of noisy data that are separated by time and space. Machine analytics are much better at seeing patterns and connecting different, disparate data sets. Robots will be able to learn at the workcell level and make changes to adjust their performance.”

Lawton predicts that the future automated factory will be continuously learning, self-configuring and self-healing. “We’re taking small steps toward that future with robots, software and artificial intelligence. Timing is the biggest question now.”


More in Robotics