A few weeks ago the Disruptive Innovators Network organised a tour at one of Amazon’s flagship fulfillment centers in Altrincham. The hype around this specific center, which predominantly warehouses smaller items, is centred on its use of robots and automated processes to manage its inventory.
Thinking about our own stock management and logistics, I tagged along to see what all the fuss is about. Here’s my three main takeaways:
Employing robotics to its natural limit
Amazon aren’t opposed to rubbing out a few job opportunities through the careful application of technology and robotics. However, on a human level, it’s quite reassuring to see people still play an important role despite all our faults - like needing the toilet and having our safety protected by law despite being so fleshy and fragile. But sentiment aside, where Amazon are forced to employ human labour is a solid barometer of the robotics industry and its current technical competency.
Consider the ‘picker’ - an Amazon employee who receives an assortment of products in black plastic crates and must distribute them in pre-assigned bins, which are brought to within arms reach from a vast warehouse by little robots. The robot disappears into the depths just as another one arrives. Wash, rinse, repeat. Why would this be so difficult to replicate robotically?
Adaptive Handling - The crates arrive directly from suppliers, who’ve found the most efficient way to deliver products (naturally) is to put an assortment in one shipment. For a human, handling a wide variety of boxes, books, DVDs and baggies is well within our means - but think about all the variables that need to be programmed for a computer to imitate this behavior. A product that’s heavy but fragile requires a totally different approach to handling than a small, flat children’s book. The technology is clearly not there yet in terms of capability or affordability.
Visual Recognition - Pickers are told which products go in which bin and, at the supply end of the process, which products need to come off a bin. Differentiating your product, in a dark cubby-hole alongside a number of other weird shapes and colours, is pretty simple for a human to do. We’ll rummage, recognize an item despite its orientation and take it for a closer look. That’s a lot of data to replicate with a machine!
Visual Quality Control - A certain threshold for damaged product is natural - but how can a computer understand what a dog ear looks like in comparison to a pristine book, or what a slightly crumpled box might look like? If you want to prevent humans initiating costly, preventable returns, you need humans to make a judgement call on what is acceptable and what isn’t.
It’s worth bearing these technological deficiencies in mind when thinking about the future design of our own services. If Amazon, who are steadfastly spurred on by efficiency, can’t totally automate their processes yet then the housing sector should temper its expectations.
“You’re only here because a robot couldn’t do it better”
The obvious implications on workplace culture are pretty sucky - and stories about Amazon’s approach to treating workers like robots are rife. Historically the company have been quite unflinching in prioritising its customer’s experiences over that of its employees, particularly at the swampy nether-regions of its organisational hierarchy. So why then run these factory tours at all - given that nothing is accidental, everything designed?
Factory ambassadors greet you at the entrance, beaming from ear to ear. Collectively you bounce to a meeting room where you’re given an introduction, high-vis and headset for your journey around the warehouse floor.
If I were cynical, I’d assume that this is an attempt to alleviate negative public image about its working environment - guised under a presentation of robotic technology and automation. In fact, the constant reflection on workplace culture and break allowances by our primary ambassador enforces this message, if only it wasn’t for the careful chaperoning, security gates on entry/exit and total absence of natural light I might have believed them.
The manual jobs are undoubtedly pretty monotonous and repetitive physically as much as mentally, so could clearly take a toll on a human operative over time. The work is often competitive, as its low skill level means there’s people lining up around the block to step into your shoes - but who’s fault is that. Amazon’s or wider societies? Picking is not the most inspirational job, no, but as long as environments are safe then we’ve not got a lot to beat Amazon up with (in my opinion). Arguably you’ll have nothing to wield when, in 5 years, robotic technology has evolved to the point it no longer needs human intervention in its fulfillment centers at all - as I’m sure Amazon will be first in the queue for such technology.
Using data to break away from convention
When you go into a supermarket, what do you notice about stock location? If you want a loaf of bread, you expect it to be next to the rest of the bread, right? There’s probably cake or crumpets or something nearby too - maybe even baking ingredients?
That’s been designed for you as a customer, both to appease your natural instincts about where to go - making your physical journey around a supermarket as short or meandering as you’d like - and to exploit opportunities to market more products to you on your journey.
But if there are no customers to speak, or the ‘journey’ has all been met digitally, the task is now to get product from A to B as efficiently as possible and we don’t need to accommodate that human intuition anymore ‘randomness’ becomes a virtue. Amazon distribute their stock as randomly across their warehouses as possible (with the marked exception being regular or popular orders being located on the peripheral of the warehouse), which gives pickers at different stations the best chance of fulfilling similar orders simultaneously.
What this represents is a paradigm shift - a seemingly unintended efficiency resulting from implementing robotics that changes the game forever. Shifting back to human warehouses not only hugely bumps up labour costs, it’ll require a complete re-calibration of stock - both stripping away precious competitive edge.
So - robots. Invariably more affordable, more efficient and much cooler than a human (...sorry humans) but also technologically limited for now, undermine the job security of entry level workers and require a completely different approach that makes it difficult to return back to a human workforce. But in a fiercely competitive marketplace the pros clearly outweigh the cons already - just don’t expect Johnny Five in any housing association distribution centers for a couple of years…