When visualizing industrial robots, the classic six-axis arm comes readily to mind. In manufacturing applications, these robots have spent decades living up to the definition of the word robot as “a device that automatically performs complicated or repetitive tasks.” While there is no shortage of complication and repetition in warehousing and distribution, robotic automation has had to develop a whole new skill set to tackle the challenges of those unstructured environments.
In addition to conventional robotic arms, solutions such as goods-to-person systems, automatic guided vehicles (AGV) and shuttle-based automated storage and retrieval systems (AS/RS) can often be bundled under the broad umbrella of robotics. In an effort to overcome variables like SKU proliferation, labor shortages and smaller, more frequent orders, both the hardware and the software of these systems have come a long way in a short time. While some solutions take humans entirely out of dull, dirty or dangerous tasks, the majority of robotic implementations are now focused not on replacing workers, but optimizing them.
“Humans are the most dynamic piece of equipment you could ever deploy,” says Adam Brown, manager of integrator partners at SSI Schaefer Systems International. “They are more capable and flexible than anything—if you can minimize movement and fatigue to keep them focused on what they do best, which is dynamic manipulation.”
While possible, it is impractical to build a mobile robot with sufficient vision, dexterity and speed to go out into a warehouse and pick an order before packing it and delivering it to the dock. “Don’t worry too much about the big, ultimate utopia of what it could be,” says Jim Lawton, chief marketing officer for Rethink Robotics. “If you look at the effortlessness with which a human can shuffle a few small items into a carton, robots are not even close.”
Instead, it is preferable to slice each application into discrete pieces to find the best place for humans, robots and related systems. This is where software comes in, granting the visibility to isolate order profiles and the flexibility to configure robots on the fly. Modern recently spoke to a few industry experts to find what they, and the robots, have learned from an assortment of present-day robotic applications.
Robots have already proven themselves in applications such as palletizing or depalletizing single SKUs or layers and moving pallet loads. While automated solutions are readily available around the “outskirts” of a facility, Brown says that things get increasingly difficult on the picking side. “There are a number of ways to automate pallet movement, but as products get smaller, the complexity is exponentially greater,” he says.
Similarly, as orders become smaller and order sequencing more critical, the complexity of a mixed-case pallet increases. Humans will tend to outperform a robot when presented with random products and tasked with stacking them. That said, software can help retrieve items in a certain order and conveyors can present them to the robot in a predictable way. The robot then works in coordination with neighboring technologies instead of serving as a catch-all like many humans do. “Parallel processes and a distributed attack to the problem are usually better than a one-shot solution,” Brown says. “It helps to break out these capabilities from the robot so it’s not doing everything: seeing, orienting and grabbing with the right end effector and depositing.”
While choreographing product flow, software can take into account all product dimensions and classifications, such as heavy items that need to be stacked below light ones. As soon as an order comes in, the finished pallet is built in virtual space before the system directs how the order will be picked in reality. Part of the order might come from manual picking and part from automation, says Brown. All the decisions about how each item is picked and when it’s picked are determined by the software package, giving the robot advance notice of the items headed its way.
Many robots, one mission
Inside a shuttle-based AS/RS, there are even more tasks to delegate as a fleet of independent robotic shuttles coordinate the fulfillment of orders while keeping an eye on one another. The system behaves like a swarm that dispatches each mission to members of the colony in the best position to serve it, according to Miguel Pinilla Burgos, vice president of global logistics solutions for Dematic. Burgos says there are two levels of intelligence; one understands the operation and one is the colony or system intelligence.
Combined, these levels of intelligence keep the operation moving despite a failure of one shuttle, a jam somewhere in the system, or during periods of planned shuttle maintenance. “Say there are 10,000 transactions per day, per shift,” explains Larry Sweet, chief technology officer for Symbotic. “If the robots are successful 99% of the time, you might say that’s a good number, but it’s also 100 issues where someone had to address something. The level of reliability and availability are very high to make these systems work.”
Even when not fulfilling orders, shuttles will perform “inventory grooming” by consolidating and optimizing stored items. This means that the system is constantly repositioning the inventory it is storing and handling to optimize throughput. If the system serves totes to a goods-to-person picking station, grooming can help improve speed and picks per retrieval. Shuttle systems also sort automatically, Sweet says, eliminating the need to sequence, sort or merge downstream. “With conveyors and sorters it might not be as easy to change a physical layout or software to accommodate changes,” he says. “Both the hardware and the software need to be flexible. In terms of software, hard-coding is not preferable to ‘learning’ systems that can dynamically add or remove SKUs.”
Robots with intelligent software can work to problem-solve independently. In a traditional top-down system, the ERP drives the WMS, which controls the WCS, which drives the equipment, Burgos says. “I think that model is likely to be greatly upset going forward,” he says. “There’s no reason replenishment signals need to go up to the ERP and down again. For instance, when a demand signal comes into the packing station, it might ask for items to be delivered in the optimal sequence for packing. This propagates demand upstream, and you see processing among equipment elements instead of centralized command and control.”
Of course, a robot does not necessarily need to communicate with anything else in the warehouse, according to Matt Wicks, vice president of controls and software for Intelligrated’s manufacturing systems division. Plenty of vision-guided robotic depalletizers are stand-alone units controlled from a nearby panel. But the benefits of communication among pieces of equipment can also break down traditional barriers between warehousing and manufacturing. “We’re trying to promote that concept with our customers,” Wicks says. “Too many think that one process ends at manufacturing and then resumes separately in a warehouse or DC. There can be a lot of efficiencies in tying them together.”
Wicks offered the example of a beverage manufacturer that identified the requirements to support distribution at the point of production. Using a robotic depalletizer, conveyors, AS/RS, a case handling and sequencing system, the warehouse ultimately served as replenishment for order fulfillment as well as a buffer for the raw materials used for manufacturing. “They stored both unfinished and finished goods within the same AS/RS,” says Wicks. “The same tasks could have been accomplished without robotics, but it would not have made financial sense. With all of the other factors that came into the justification, robotics were critical to meeting their needs.”
Robots as employees
Many justifications for investment in robotics center on labor, but it is rarely possible to simply replace a human worker with a robot. Besides, Sweet says, “it’s not just substituting automation for a human, it’s about using it to make the human more productive.” When breaking down a process into steps suited for one or the other, it is sometimes possible to surround a conventional robot with guarding and create some sort of hand-off between humans and automation. The downside is that the cage and robot can become a monument, a fixed installation around which everything else must adapt.
This is where a new generation of collaborative robot systems offers the flexibility to change position and performance on the fly. “Mass manufacturers have the ROI equation easily, but that model—one robot, one place, one task—doesn’t work for a lot of small- and medium-sized businesses,” says Ed Mullen, national sales manager, North America, for Universal Robotics. “It’s the same concept with an employee. You want him to learn a number of things, be versatile, grow, adapt and prove his value.”
Lawton at Rethink Robotics agrees, suggesting customers want to be able to ask a robot as they would an employee: “What have you been up to today? What were some challenges? What do you plan to do tomorrow?” Unlike integrated robotics dependent on specialists for programming, collaborative robots are designed to accept intuitive and gesture-based commands while operating safely in close proximity to human counterparts. Just like a human, they are also expected to “learn” a certain amount of self-sufficiency, for instance by retrieving a dropped object rather than requesting human intervention.
Some collaborative robots are accurate to four thousandths of an inch and can integrate with vision, networks or PLCs. But a robot that an average worker can use the day after it is shipped to the facility is not a system, Mullen argues, but a tool akin to a power drill. The concept seems to have some appeal; after shipping 31 robots in 2009, Mullen says his company now ships 130 per month.
Perhaps one of the biggest benefits of a robotic solution is that robots can collect data as they complete tasks. Lawton describes a customer who needed cycle time data, handed clipboards to operators and collected a lot of essentially useless information. “With a robot maintaining telemetry about tasks, you can use that data to improve the quality of the process,” Lawton says. “That’s insight you wouldn’t be able to gather otherwise.”
Companies mentioned in this article
Rethink Robotics: rethinkrobotics.com
SSI Schaefer Systems International: ssi-schaefer.us
Universal Robots: universal-robots.com
Is there a robot in your distribution future?
Genco is looking to piece handling robots and automatic lift trucks to take its business to the next level of materials handling.