Rise of the Robots - Lessons in Automation from Amazon
As someone who grew up in the Seventies, the term robot always evokes fond memories; the bleeping or whistling of a companion like R2D2, almost canine-esque in its loyalty to its human master. However, what once felt like science fiction from a galaxy far, far away is now starting to become reality and continuing to change the balance between man and machine in so many parts of our economy.
With that in mind I decided to take a Friday afternoon trip to tour the Amazon Logistics centre to see how our biggest online retailer is making use of these technical advances to streamline its supply chain and logistics processes. I should probably explain at this juncture that seeing how the warehousing & distribution process has changed was of particular interest to me as I spent every summer as a student working in a music warehouse in West London. It therefore provided a great comparison as to how the various stages of this process (goods in, picking, packing, goods out) have changed over the last 20 years.
The first thing to say is that watching a floor of robots buzzing around moving shelves in perfect harmony is actually quite transfixing, although the five year old Star Wars fan in me was quite disappointed they don’t bleep and bloop like R2! On a serious note though what’s fascinating is that the robots have enabled the process of picking orders to be completely flipped on its head. Twenty years ago, we walked up and down the aisles of the warehouse, packing orders into a box on a trolley like a supermarket shop. Now the pickers remain at their station with the robots bringing the right shelf to them at the right time to fulfil the order.
Robots at Work at Amazon Fulfilment Centre
Secondly, there are very few parts of the process which are completely automated; for the most part machines are assisting the process and making it quicker, faster and easier rather than eliminating human input completely. The one exception to this is Amazon’s SLAM machine (scan, label, apply, mail). Twenty years ago , four of us would be running up and down the line applying the right coloured sticker to the right box so it went on the correct palette for each region. This was fantastic at teaching you the geography of the UK and a wonder to watch it perfected into an art form as the most experienced workers were able to buzz up and down the line effortlessly with a sticker on all 10 digits and apply them to each box in an instant. However, even for these sultans of stick, the process was slow and prone to error whereas the SLAM machine is effortless and error free (see video below at 2:00).
End to End Process at Amazon Fulfilment
Perhaps most importantly this stark comparison of the changes in the warehousing process over the past two decades made me reflect on the changes in our own industry. Much has changed in the world of research and similar efficiencies have been achieved; data arrives in near to real time and we don’t wait for the tome of paper tables to emerge from DP. However, on the other hand, it’s hard for us to conceive how the changes in technology will shape our roles in the future. For the most part this is because our tendency is to overlay tomorrow’s technology onto today’s processes. Inevitably there will be areas where the whole process can be flipped on it’s head – think bring the shelves to the picker rather than the picker going to the shelves.
Reversing the picking process in Amazon warehouses allows Amazon to store products randomly in its warehouse rather than in a structured, logical categorisation. If it’s not stretching the analogy too far, similar advances are happening in the data structures which we now rely on. We can store more information in an even less structured way yet increased computing power enables the user to make decisions off the back of that data.
The impact of AI was much discussed at the MRS Conference this year and I was pleased that, much like the Amazon Fulfilment Centre, the vision painted was one of augmented intelligence rather than a head on collision between man and machine. I loved Stan Sthanunathan’s quote:-
"AI can help us augment human intelligence rather than replace human intelligence – make machines work so that humans can think"
Sectors like logistics are more straightforward to automate because speed and accuracy rather than intuition and creativity are of the essence However, if we’re honest with ourselves, there is still a huge amount of manual heavy lifting throughout the research process and the core steps in the journey remain, in some areas at least, the same as they were ten even twenty years.
It does though feel like we’re at a tipping point and technology is finally enabling us to not only to do things quicker and faster but also to tap into rich sources of insight (such as text and imagery) that were previously beyond our reach at scale. Of course there is a need to weed out the snake oil sellers and also to ensure appropriate ethical standards are in place but it’s critical for us to lean in as an industry and embrace these new approaches rather than being fearful of the impact on the roles of today…