RoboBusiness Leadership Summit: Future of robotics unveiled

The 8th annual RoboBusiness Leadership Summit (RBLS) drew 400 attendees and 35 exhibiting sponsors to Pittsburgh, Pa., in October to unveil new technologies, exchange best practices and explore opportunities to bring robotics from the laboratory to industry.

Produced by Robotics Trends, a division of EH Publishing, the event showcased some of the top minds in robotics and offered bold predictions for the future. The event also illustrated robotic solutions at work in warehousing and distribution, from technologies that can be deployed within a day, to grocery warehouses where humans never touch product.

But for all the recent achievements on display at the event, the prevailing tone was more of a “wake-up call” to industry. Having identified industrial spaces and warehousing as key opportunities for growth, presenters and attendees were keen to address the clear obstacles to adoption in those notoriously tech-wary customer bases. Combating the perception of robots as inflexible, costly job-killers was at the top of the agenda.

Over the next pages, Modern has assembled some of the warehousing and distribution highlights from the event. More than a few of the presenters mentioned Amazon’s acquisition of Kiva Systems, a move that had clearly stirred up the robotics industry. As a model for the future expansion of robotics in warehousing, Kiva’s solution proved two things: Robots and humans have complementary strengths (moving goods and dexterous grabbing, respectively), and sometimes the simplest approach is best.

Session: Industrial robot solutions for warehousing and distribution operations
In December 2010, a fully automated warehouse owned by C&S Wholesale Grocers
went into operation in Newburgh, N.Y. (See case study, page 90) The previously conventional facility, which serves 146 supermarkets, had been retrofit with robotic depalletizers, a case-based automated storage and retrieval system (AS/RS) and robotic palletizers. After a worker scans the pallet at receiving and removes the shrink wrap, no human touches product until it arrives at the stores.

During his presentation at RBLS, Larry Sweet, chief technology officer at Symbotic, which is a member of the Material Handling Industry of Americas (MHIA) AS/RS industry group, was joined by project collaborators from automation designer and manufacturer Hartness International and FANUC Robotics. Sweet explained why the Newburgh facility represents “disruptive” technology—or that which reimagines a solution instead of iterating on previous versions.

“This is not your grandfather’s AS/RS,” said Sweet. “This is disruptive technology that’s going to cause an explosion in the marketplace, both for mobile robots and conventional robots.”
The key to this disruption, Sweet said, is the central storage and order fulfillment engine, which uses high-speed autonomous mobile robots (AMR) to store and retrieve cases in a 3D storage space. Without fixed storage locations, each case takes up only as much space as is needed for the robots to access it.

After pallets are received, robotic depalletizers place cases onto a conveyor system designed for singulation. This technique ensures that there is space between each case for robots to differentiate them and for grippers to operate. At this point, the AMR “rovers” place each case into randomized storage locations that can be optimized on the fly.

“Now you’re talking about planning in three dimensions,” said Sweet. “Traditionally, AS/RS are columns, whether pallet columns or case columns. This system behaves more like a computer’s hard drive, optimizing all available space.”

Like a computer, the system relies on complex software to coordinate the fulfillment of orders. Before the first case of an order is picked, software calculates the makeup of each pallet, producing stable, store-ready pallets as much as 8 feet high. A human selector picks 150 cases per hour and is solving the three-dimensional stacking puzzle as he selects, said Sweet. “This facility ships 20 million cases per week, so when you do the math you can understand the business reason behind this,” he said. “This is a market that is screaming out for a solution.”

About 95% of the facility’s 6,000 SKUs are compatible with the system. Any SKUs that are oddly shaped, such as broom sticks and bags of dog food, or for some reason will not be accepted by the robotic palletizer, are automatically routed to a manual station. A single palletizing robot has a 6-second cycle time, for total of 600 cases per hour, as opposed to the human’s 150. The system also eliminates the need for 3% of pallets to be manually audited.

The most difficult and most critical hurdle is sequencing, said Sweet. “Anyone in this industry who has tried to implement traditional fixed automation will find they face a choice between sequencing and throughput.” This system achieves both, sequencing each case for store aisles, and each pallet for loading onto trucks. “Past solutions have had some success with low-SKU pallets, but struggle achieving the complexity common to many grocery orders.”

Traditional automation can cost as much as $100 million since they tend to require a new building, Sweet said. Operating costs are also high, with highly trained technicians necessary to service as many as 10,000 motors or more.

“But once you’ve built all that physical hardware, you’re pretty much stuck with it. If you incorporate mobile robots into that equation, basically everything I just said that was a negative turns into a positive.” C&S enjoyed a low cost to build, short installation time, low operating costs, and easy adaptation to business fluctuations. Labor savings, workers compensation costs, product damage, and transportation costs have all improved.

Session: Machines and man working side by side
Rodney Brooks, founder of iRobot and co-founder, chairman and CTO of Rethink Robotics, emphasized a near future where employers will not consider robots and humans as alternatives to one another, but as coworkers.

“Look at any task. There are things a human contributes that a robot will not be able to do for a long time,” said Brooks during his session at RBLS. “For instance, Kiva recognized that robots move goods very well, and humans do dexterous grabbing very well. The result is a very valuable pairing.”

Brooks said he is now focused on developing industrial robots to work intimately with their human counterparts, with safety measures, interfaces and learning capabilities designed for easy use and integration. With the planned January release of the humanoid robot Baxter, Brooks’ aim is to make the machine productive in its first day on the line.

“Think more iPhone, less mainframe. The robot is the interface,” said Brooks. Baxter requires no system integration and no program writing. The unit is “teachable,” said Brooks, who then demonstrated how it could be taught to blind pick from one area and place in another in less than 30 seconds. Because it is teachable and does not rely on sequential programming, the robot can adapt to tasks as they are presented. For instance, the machine might know how to perform steps one through 10 of a process. If a human should suddenly perform steps one through five, the robot will evaluate the item it is handed and skip to step six.

A customer could even design customized end-or-arm tooling that the robot would “learn” how to use without requiring it to be programmed with the tooling dimensions. It requires no network to operate, and any learned behavior from one robot can be transferred by thumb drive to any other robot.

“Ease of integration is essential to smaller operations and those unfamiliar with integrating robotics systems,” said Brooks. More importantly, Brooks sees this movement as essential to creating and sustaining a robust manufacturing workforce in the United States.

“The robot is a hammer, and many have built their line around making a nail for the hammer to hit,” he said. “To a large extent, that’s still the way it’s done.” With this approach, a robot can jump into a larger variety of applications with minimal disruption. There are 300,000 small manufacturing companies (fewer than 500 employees) in the United States, and almost none of them use robots, said Brooks. That will change, he said, as robotics manufacturers work to create solutions that adapt to the application, instead of the other way around.

“Don’t create new standalone technology,” Brooks advised his peers. “Listen to the customers’ needs and then build to meet that need and solve a problem.”

Session: Mobile service robots in manufacturing
Autonomous mobile robots (AMR) allow end-users to solve problems that could not be solved in the past, said John Dulchinos, president and CEO of Adept Technologies. Adept’s AMRs compete primarily against conveyors and movement by hand, but those using conveyors to automate material flow are less likely to rip out the conveyors in favor of AMR. Applications with product movement by hand for which there have not historically been any viable alternatives will now be appealing candidates, he said.

Dulchinos predicted the end of the centuries-old pursuit for low-cost labor. “Demographics and technology will conspire to end the hunt for low-cost labor while spurring robotics growth,” Dulchinos said. He presented data indicating that by 2018, 70 million baby boomers will retire and 40 million new workers will enter the workforce, creating a gap of 30 million workers.

“In 2,700 years, one of the biggest—and one of the only—major leaps in materials handling is the transition from moving items to tracking them while they move,” he said, emphasizing that AMRs inherently provide real-time traceability. “In materials handling, there are two segments. One is people, with pushcarts, forklifts, or just carrying things, who are very flexible but not very consistent. The other segment is conveyors, AGVs, and other devices which are inflexible, but very consistent. We need to bring the two together. Robots will not be replacing people, but functions.”

Dulchinos said there are two main challenges to bringing these two halves together. The first centers around mapping and autonomous navigation, those technologies that allow robots to move confidently and safely around people in a crowded environment. The second half is the coordination of a fleet of robots with real-time business imperatives. Dulchinos said e-commerce and home delivery applications will be good candidates for AMR, as well as traditional warehouses in need of more flexibility.

Session: Robots in the workplace
RBLS featured a strong emphasis on mobile autonomous robots. This technology has already made its presence known in warehousing and distribution, where solutions from folks like Kiva have had a big impact.

However, robots need not require a warehouse to redesign processes and product movement from scratch. They also don’t need to be fully autonomous. A growing line of automated lift trucks is aimed at reducing the amount of labor needed to perform non-value-added tasks, such as ferrying loaded or empty pallet jacks from place to place. As opposed to the humanoid “coworkers” on display at RBLS, these robots behave more like smart tools. They deploy fast, require minimal operator training, and accept direction only from operator input.

In Giant Eagle’s 400,000-square-foot DC in Pittsburgh, Pa., one operator is responsible for directing a fleet of five semi-autonomous pallet jacks. An operator can teach the unit a number of routes, which it remembers instantly through 360-degree vision and odometry. Using no lasers, magnets, or external guidance of any kind, the units photograph their environments thousands of times in their travels. This allows them to keep their bearings to within one centimeter, even as product faces and other warehouse elements move and change.

Robert Kuchta is the distribution systems manager at Giant Eagle, which has served as a test bed for Seegrid automated lift trucks for the past five years. In that time, the units have evolved from experiments to trusted and valued partners in the facility.

“There was some concern in the beginning, but it didn’t take very long at all before everyone saw the value of the robotic trucks,” said Kuchta. “Now, we can’t imagine working without them.”
In addition to automated lift trucks purpose-built by Seegrid, the facility is testing Raymond trucks equipped with Seegrid vision-sensing units. With a flawless safety record, these units help the 8,000-SKU facility move more than 700,000 cases per week. The Raymond/Seegrid product will likely launch in 2013, according to Michael Hasco, chief growth officer at Seegrid.

In his presentation, Hasco was joined by Larry Baldauf, retired senior vice president of supply chain and logistics for Giant Eagle. Baldauf described the more than decade-long relationship he had formed with experts at neighboring Carnegie Mellon who went on to found Seegrid in 2003.

“In the mid-90s, we had engineered labor standards at Giant Eagle,” said Baldauf. “We knew where the product was, we knew where the people were, and we were able to squeeze costs down as far as they would go while still being fair to our people.” Having picked most of the low-hanging fruit, Baldauf began thinking creatively about reducing non-value-added movement. “I went to CMU and said, ‘You guys can put robots on Mars. You can’t tell me you don’t have something that can move product from point A to point B in a warehouse.’”

After eight months, the collaboration had produced a pallet jack that could move around by looking at the “fingerprint” of the warehouse floor. Having served as a beta site for the Seegrid technology since its inception, the facility has shaped the current version into a rugged, warehouse-proven unit. “My job as a beta site is to figure out how to break it, how to push it to its limits,” said Baldauf. “In all those years, no Seegrid unit has hit a person or a part of the building, ever.”

Offering a 9-month ROI, the unit can start working and start saving money within two hours of its delivery to the site, Baldauf said. The truck costs $3 per hour to operate as opposed to $25 to $30 per hour for a manned unit. “About 73% of your operating costs are labor. You’ve got to attack that labor number,” said Baldauf, who said the technology has not resulted in any lost jobs. “In fact, we’re now getting more done with the same people.”