Share |

Focus: Distribution/Materials Handling

Feature Article from Our Distribution and Materials Handling Subject Area - See All

From SCDigest's On-Target E-Magazine

- July 14, 2015 -

Supply Chain News: Are Collaborative Robots Ready for Prime Time in the DC?

We Say the Jury is Still Clearly Out, though there Appears to be Many Applications in Manufacturing

SCDigest Editorial Staff

Are so called collaborative robots, of which the Baxter machine from Rethink Robotics is the best known example, ready to play key roles in many distribution centers any time soon?

After seeing a presentation at the WERC conference in Orlando in May from Chris Harbert, global sales director at Rethink, we'd say the jury is still out, though there appear to be many applications in manufacturing.

There was certainly a lot of interest in the topic, with a very packed audience for Harbert's session. Rethink was founded in 2008 by Rodney Brooks, formerly of head of the artificial intelligence lab at MIT and co-inventor of the Roomba robotic vacuum. The company's Baxter robot - about six foot tall, with a face, two arms, and a vison system, was released in 2008.

SCDigest Says:

There were other application examples of simply moving items from one place and putting them in another, generally in manufacturing. Kitting applications of various types were cited but not detailed.

What Do You Say?

Click Here to Send Us Your Comments

Click Here to Post or See Reader Feedback

Harbert started by noting a number challenges in distribution, including a real labor shortage in many areas, rising labor costs, and the difficulty retaining good associates. What's more, the millennial generation overall seems little interested in jobs inside a DC. Currently, 40% of US warehouse workers are age 45 or older - and that percentage keeps growing

That would seem like a situation ripe to automate with robotics, but traditional industrial robots often don't have a good fit in a DC environment. They work best when they can automate a specific task millions of times over a period of years. But they are inflexible, expensive to buy and especially deploy, and dangerous, usually walled off in some way from human workers.

"And if you need to change something, that's a hard day," Harbert said. "Most robotic automation is designed so that you set it up, get it working, and then you don't touch it."

But DCs don't generally present themselves with opportunities for that kind of robotic automation, though certainly there is some movement towards robotic case picking and palletization. Conversely, while humans are very versatile, they are increasingly expensive.

So a new generation of what are being called collaborative robots was born, offered not only by Rethink but also Kuka, Universal Robots, ABB and others, meant to fill the gap between human-only processes, and robot-only processes.

A key feature of these new machines is that they are very safe, and can work right alongside of humans. They are inexpensive compared with traditional robots, and can be trained to do work by showing them what needs to be done, moving their arms and having the robot grab an item, for instance, rather than expensive programming by system engineers.

"We need to enable people to do what they are best at and take some of the repetitive work, the dull, the boring, the dirty work, and give that to robots," Harbert said.

He said traditional robots might average $130,000 or so each to install, but it turns out the hardware itself is not the biggest cost element, representing maybe only 25% of the total cost There are usually large costs to acquire all the gear needed to protect workers from the robot and especially programming and field implementation by expensive specialists. Instead, a fully loaded Baxter can be purchased for only about $40,000, for example.

Another goal with collaborative robots is to dramatically reduce the investment to get a robot deployed, with the non-programming approach to teaching such a robot what to do a key aspect of those savings.

In fact, most of the on-going development for Baxter and other such robots takes place with the robot companies developing new what you might call a series "meta-tasks" that enable customers to train the robots to do new types of work. That approach also means a collaborative robot can generally be fully deployed for a task in three to as little as just one day, versus multiple weeks for traditional robots, using with general purpose workers or managers versus specialized systems engineers.

Most companies designate one of more people as "super users" to do all the robot training, Harbert said.

What Else Makes Collaborative Robots Different?

A number of other capabilities make collaborative robots different. Rethink and others, for example are trying to add "common sense," or what Hartbert refers to as behavior-based intelligence, into these machines. What does that mean? If the robot senses it is getting into trouble, for example, based on the work coming at it, it should stop and ask for help - similar to what a human would do.

These new age robots are generally also easily movable - not anchored down like a traditional robot, and someday soon may even be able to move themselves to a new position. That means they can easily be moved to do the same task they were doing in a different area of the plant or DC, or do a completely different task in a new location.

Harbert gave an interesting example of that positionability of the Rethink Robots. One manufacturer would move Baxters around during the day, and found it had to spend a lot of time to reposition the robot when it was brought back to a given work cell to exactly where it had been previously, because all the coordinates needed for the task were based on that specific position.

Rethink solved that probable by installing fixed "landmarks" (2D bar codes) at each cell, and then using Baxter's vision system (also a characteristic of collaborative robots) to pick up the landmark and thus know where it was in relation to that fixed position. All the other coordinates were then adjusted based on where Baxter now knew it was standing. Set up now just takes a couple of minutes when a robot is returned to a previous work cell.

The vision systems referenced above can help provides context to the robot. It can "see" if a given part is where it is supposed to be, for example.

By contrast, most traditional robots only know position, coordinates and trajectory - very precisely where and in what path they are programmed to grab and move some item. But if the item is misplaced even a little, robot will still move its arm to where it is programmed to go and miss the item. It will usually continue to move, because it doesn't even know it missed the item.

Another difference is that collaborative robots generally have a full seven degrees of freedom for each robotic arm, providing the agility needed for most tasks.

Harbert also said the goal is also to provide a fully integrated system, with all capabilities needed coming with the robot --a robot, any required peripherals, the controller, vision system, etc.- all in one package, which will also reduce implementation time and costs.

(Distribution/Materials Handling Story Continues Below )

CATEGORY SPONSOR: LONGBOW ADVANTAGE - JDA SUPPLY CHAIN CONSULTANTS

Learn More about Longbow's

Expert Services for JDA WMS

Download Longbow Advantage

Business Briefs

The Keys to WMS Success,

Maximizing JDA WMS

Performance and More

As far as safety and the concept that these new collaborative robots are designed for interaction with humans, Harnert showed interesting video of a himself working right next to a Baxter robot and purposely getting occasionally hit with the robot's arm, which delivers a very software impact and immediate recoil that has almost no effect on him, after which he then grabs the robot arm and easily moves it to the other side of his body.

After the collision, the Baxter stops for a few seconds, but then finished the job of placing an item in a box.

An interesting discussion around safety then ensued based on a question about whether these new age robots met OSHA standards.

Harbert said OSHA rules for traditional robots were driven from existing RA/ANSI standards for industrial robots that were developed back in 2008, before these new collaborative robots were developed. So, they don't really fit for these much safer machines.

He said an ANSI committee has been meeting to develop a new set of standards specifically for collaborative robots, which are expected later this year.

"I have had a few OSHA regulators that have been concerned, but we've never had any problem with an implementation," Herbert said. "It is something of a hassle today because you can't just check some boxes and be done yet before the new rules are established, but again we've never seen an implementation stopped."

From a safety technology perspective, he said that while some collaborative robots use internal breaking systems, Rethink Robotics does not, using instead internal springs in the joints that act as shock absorbers, plus torque sensors in every joint that feel the force of an impact and tell the robot to stop movement. He noted that if any of the robot's sensors fails, the control system will shut the robot down.

We'll note, however, that this safety feature comes with some inherent downside: safety is achieved though moving relatively light "payloads" at fairly slow speeds. Increasing the weight handled or the speed of movement - as might be needed or at least desirable for some tasks - would then create some of the same safety risks as traditional robots.

Baxter robots and other also have tactile sensing in their "hands." That helps deal with misplaced items, with Harbert showing a video example of a robot successfully moving a piece of glass to another position even though it is off center by using its tactile sensing capabilities. Those capabilities could also be used to ensure an item was placed correctly into some kind of packaging, for example, perhaps by feeling a given level of resistance before it let an item go.

Compare that to traditional robots, where items typically can be no more than 0.2 millimeters from the programmed position point. That also means that with collaborative robots some companies would have to spend less money in systems needed to precisely position items presented to the robot.

What are the Applications?

Harbert showed videos of several different real life applications. One, for example, was what he called "line unloading" of a manufacturing line producing a part that was not made very often. Here, the robot simply takes parts off conveyor and places them into tote bins. Before, the factory had to move someone off a full time production line to come and manage this process for maybe six hours every week or so.

And this, Harbert said, illustrated a key point: "A lot of the use examples don't call for super high speeds."

Another example involved metal machined parts that had to undergo some sort of washing process. Here, a Baxter takes the part and places it in the washer machine, closes the door, waits patiently there while the part is processed, opens the door, takes the part out when finished and places it into a tote. Before, a human stood there and did this all day.

There were other application examples of simply moving items from one place and putting them in another, generally in manufacturing. Kitting applications of various types were cited but not detailed.

SCDigest's take: There seem to many applications for these types of robots inside a factory, especially for the highly repetitive, boring jobs few want to do anyways. In fact, Rethink has taken the lead in suggesting collaborative robots could play a key role in the "reshoring" movement by reducing labor costs for these low value added jobs at US factories.

There may be some of these types of tasks in distribution centers too., In fact, at another conference earlier this year a manager from 3PL Genco said it was using a few Baxters to take solid SKU cartons of DVD titles, unload them and placing the DVDs on a takeaway line so they can be shipped to customer sites at less than full carton quantities. But these types of slower moving tasks are less prevalent in distribution than manufacturing, we think.

Our sense is that Baxter and his new buddy Sawyer from Rethink are still awaiting a "killer app" for distribution.

A recording of this WERC presentation is available here: Real Robots in the Warehouse

What do you think is the future for collaborative robots like Baxter in the DC? What could be the killer app? Let us know your thoughts at the Feedback button (email) or section (web form) below.

Recent Feedback

No Feedback on this article yet