Amazon Prime Air Will Fly on Big Data’s Wings – by Beth Schultz, Editor in Chief, AllAnalytics.com
Amazon, one day in the not-so-distant future, wants to set the air abuzz with package-delivery drones. Today, it will have to be satisfied with buzz of a more personal sort, as industry watchers analyze the feasibility of the proposed Jetson-like Amazon Prime Air service.
Should anyone have missed this news, let me recap. In a 60 Minutes interview that aired Dec. 1, Amazon CEO Jeff Bezos described his vision for a world in which you can order that to-die-for sweater you’ve just seen on Amazon and be wearing it 30 minutes later. Amazon would use mini-drones, called octocopters for the eight blades they sport, to deliver the package to your doorstep — optimistically, as early as 2017.
The plan comes with caveats, to be sure. For one, you’d have to live in a metropolitan area near an Amazon fulfillment center — but the idea is fascinating nonetheless. As 60 Minutes reporter Charlie Rose said during the program, Amazon wants “to sell everything to everybody around the world, as fast as possible.”
Bezos readily admits that much work remains to make mini-drone deliveries on a large, commercial scale possible, and skepticism is certainly rampant. But we already know the power of prescriptive modeling and optimization in fine-tuning a global package delivery network, from our conversations with Jack Levis, director of process management at UPS. (See Inside Analytics: UPS Delivers the Goods.) So maybe the idea isn’t quite so impossible after all — algorithmically speaking and from a data perspective, at least.
For insight on that, we turned to Roei Ganzarski, COO and president of BoldIQ, which had powered the now-defunct (but not for lack of optimization) DayJet on-demand air taxi service. BoldIQ is in use within a number of other industries that have real-time needs and lots of disruptions with which to contend. Ganzarski brings his experience as a former Boeing executive to the discussion, as well.
He’s among those who say we can count on Amazon to make this happen, “as funky an idea as it is.” And in so doing, it will conquer four distinct worlds: same-day logistics; drone operations; aviation, with the introduction of unmanned aerial vehicles into airspace; and big data, which is a new world in its own right but also encapsulates the three others, he told us.
So let’s zero in on the big data nature of the Prime Air concept, as did Ganzarski: All this is going to be happening in real-time, disrupting all those worlds, each of which produces huge amounts of data and requires huge amounts of compute power. It’s not doable, of course, without advanced optimization software.
Amazon has to find a way to know in real-time as every package request comes in whether it has a drone to take it on, Ganzarkski says. If no, it’ll put it on a truck and push out delivery time to a day or more. If yes, it needs to know all the ripple effects of that decision.
He continued: If Drone No. 1 is taking this package, once it arrives at its destination, what is its next path going to be, and its next one after that? And as things change, like a new demand comes in, or a drone breaks down, or a wind burst comes in and changes the weather pattern in that urban area, what’s the next move? Amazon will have to be able to immediately — and I’m talking milliseconds — re-network and re-schedule what its drones are doing with the packages… and at the same time, let the customers know what to expect.
There is a precedent, of a sort, in on-demand private aviation, he notes. In this market, an aircraft could be at any one of the 5,000 or so airports that accept private aviation. The service also requires a pilot and copilot who are not only certified to fly that aircraft, but who also have enough rest time as specified by the FAA. Demand, generally from wealthy individuals and corporate executives, can come at any time and require a flight to and from any of those airports.
Which aircraft is selected depends on the number of passengers and their luggage: There’s a lot that goes into the matching between the right — not just the one that can — aircraft to fly them with the right pilot and copilot so the entire network meshes in a seamless fashion, so the customer gets what he wants, and the operator is able to serve that demand while being profitable and safe.
So the beginning stages of the full-on capability, whether via a software package from a vendor such as BoldIQ or developed internally at a company like UPS or, undoubtedly, Amazon, are out there and operational already. The big data piece won’t be the problem, it seems. What will be, as you see it?