In the next few years, expect the need for data scientists in logistics to continue rising steadily. “We have data running out of our ears,” said DB Schenker’s Stikes. As a 2018 article from Forbes noted, about 90% of all data that’s ever been created in the world was created in just the previous two years, and the industry is doing its best to keep up.
“Moore’s Law is alive and well in the robotics industry,” Stikes said, referring to the common observation that integrated circuitry in computers become twice as fast every two years.
So, what can we expect to do with this immense amount of data? If it’s crunched the right way, and the machine-learning functions have worked, we can accurately predict the future, some of the 3PLs said, and plan accordingly.
“If you can imagine we have e-commerce customers that are selling cell phones, and we know that whenever somebody buys this brand of cell phone, they always buy the headset with it,” said Harik. By knowing this, he added, XPO can position certain stock-keeping units (SKUs) that go together, in close proximity in the warehouse.
“The key thing about these robots is that they are going to not only line up multiple shelves of products in front of the pickers but put them in sequence,” he added. “So it’s not just you, as a picker, taking a product from Shelf A and then you wait for the robot #2 to go get Shelf B. They would all be lined up in front of you. So the intelligence is more how you sequence it all – especially through e-commerce, where the customer expects next-day shipment.”
Stikes, however, challenges the notion that warehouse automation is mostly an e-commerce concern. “The process in a warehouse is not fundamentally different whether I’m shipping to stores or to individual homes,” he said. “Industrial clients are increasingly looking at how they can better build kits. How do we build rainbow assortments going to retailers so you get to the point where you can eliminate the back room in a store?”
This is where advanced machine learning and A.I. comes in, Stikes continued. “Ýou can begin to goal-shift and then change what the system is looking for and search for anomalies.”
For instance, you can teach the system to look for traffic jams in the warehouse. If two SKUs are too popular in one location, you separate them across two different aisles to improve product flow. Or, he added, if you see that “on the third Tuesday of the fourth month,” you’re going to move a lot of a particular item, the automated systems can rearrange the warehouse for those items before the next shift begins.
Eventually, Stikes said, with enough trending data collected from these AMRs, logistics companies will be able to make fundamental changes in now warehouses are managed, organized and restocked, and even how the footprint is designed to meet the needs of the customers.
Not bad for an army of little silicon boxes.