NextGen Supply Chain Interview: Steve Banker

NextGen Supply Chain: Automation comes in many different forms these days. How would you break it down?

Banker: There are really two layers to automation. One is machine automation and the other is software automation. Both are well established in their traditional forms. For software, think enterprise systems and warehouse management systems, to name two. For machines, think robots and automated storage among many others.

Both have removed people from many processes. But they automate those processes in non-adaptive ways. In other words, they do the same thing again and again without adapting to changes in the supply chain. But that is beginning to change.

NextGen Supply Chain: What's an example here?

Banker: Let's start with software.
Artificial intelligence and machine learning figure prominently. AI is actually allowing software to review the assumptions used to manage steps in the supply chain. And as those assumptions change so does how the software reacts to the conditions.

Take software planning. You inherently have certain assumptions. Inventory is set at a particular level, based in part on lead times. But lead times can change all the time for all sorts of reasons as suppliers' and lead carriers' performance levels change. Similarly, priorities change about which customers are most important if there is not enough supply to meet demand. The supply chain is not static.

By monitoring these parameters with the Internet of Things and other devices, these changes can be closely tracked. Mix in AI algorithms and suddenly planning software can adapt to new conditions.

NextGen Supply Chain: How far along is all of this?

Banker: There are suppliers that have productized solutions today. Terra Technology, now part of E2open, has been using machine learning for demand management since 2004. Interestingly, for years they did not want to admit it was a machine learning because that took humans out of the equation and made this a black box solution.

But now machine learning is hot and many suppliers are working to develop solutions. But many of them are science experiments. I expect that in two to three years, we will see several more suppliers with productized solutions. Right now, the rhetoric is a little bit ahead of the reality. But once the machine learning algorithms team up with machine automation, there's quite a bit of potential for the supply chain.

NextGen Supply Chain: What's happening on the machine side?

Banker: I've been closely watching what's happening in robots. And they aren't the kind of robots we've traditionally found in industry.

What we're talking about here are autonomous mobile robots in distribution. The focus now is not on bolt-in-place robots in cages on the shop floor that could seriously hurt people if the two got too close. Instead, we're talking about robots that work alongside people.

NextGen Supply Chain: How far along is this? Are the rhetoric and reality closer than in software?

Banker: The two are much more closely aligned with autonomous mobile robots. In fact, the first ones here were the bots that picked up shelves and brought them to pickers – goods to picker. Kiva developed that. It was a good 10 years ahead of the mainstream. Then Amazon bought Kiva for $750 million or so and took it all private. Interestingly enough, removing Kiva from the mainstream probably moved the technology ahead.

NextGen Supply Chain: How so?

Banker: Amazon's move confirmed to people that autonomous mobile robots had good ROI. But Kiva was limited. It only brought goods to pickers. What if future robots could do the picking too? Now that's really something.

Ultimately, these eliminate people and reduce costs. And just as the software automation I talked about is trying to get more flexible, these robots are becoming more flexible, able to go anywhere in the DC and pick any product, when outfitted correctly that is. What a boost to productivity.

NextGen Supply Chain: Can you give us an example of this?

Banker: The one I've been following is at Rochester Drug. IAM Robotics designed a piece-picking mobile robot specifically for Rochester Drug. It was featured in the July issue of Supply Chain Management Review. And I'm doing a panel on the application at our upcoming ARC Forum: Digitizing and Securing Industry, Infrastructure, and Cities.

NextGen Supply Chain: So this is far enough along to really talk about?

Banker: It is. These robots roam up and down Rochester Drug's aisles, picking product. Now they are still in the learning stages here. But I expect this to move along rapidly.

NextGen Supply Chain: It seems that often our expectations are fairly high for a visionary technology like autonomous mobile robots? What's different here?

Banker: To start, the idea originally embodied in Kiva has been actively working in DCs for more than 10 years. So this isn't starting from ground zero exactly. But these robots are going to take off not just because of the software that directs them. Perhaps more important is the componentization of the robotic parts themselves. Once this happens, these will take off. It will be just like when componentization of PC parts happened.

NextGen Supply Chain: So just how quickly do you expect this to happen?

Banker: I fully expect autonomous mobile robots to mature very rapidly and be a $100 million business in 3 years. There will be several suppliers here too. It will not be confined to IAM Robotics. Furthermore, it will have a significant impact on DC staffing levels long before autonomous vehicles will ever replace drivers of over-the-road trucks.

NextGen Supply Chain: That's pretty heady for supply chain automation. We should all stay tuned.

SC
MR