It’s no secret that Amazon is one the biggest tech successes of our time. While recently controversial, the financial success of the online megastore is still indisputable. Amazon’s growth can be attributed to its obsessive focus on quick delivery and super-streamlined logistics, which was only made possible by a fervent dedication to cloud computing.
Using AWS and other integrated platforms, Amazon’s smart warehouses led to an annual revenue in excess of $386 billion in 2020. Keep reading to see just how they achieved that record number.
Deep Machine Learning AI
- The first stage in this logistical marvel starts well before a customer places an order using the AWS “forecasting system.” Using data gathered about the customer’s location, interests, and purchase history, algorithms predict what items are needed where and when. Predicting demand is crucial to getting items one step closer to the customer… before they’re even needed.
- Amazon then uses its own platforms Aurora and Neptune to manage inventory history, transactions, and database needs in the warehouses themselves. Items are sorted randomly, but carefully catalogued by AI and picked by staff.
- The smart warehouses also use AI throughout the whole process from stocking to delivering, determining weights of bins, similarly-looking items, where an item is placed, and what the item is. When not confident in an item, the picture is sent for verification and the program continues learning about every item stocked by Amazon.
- Machine learning also determines what box to use, how much tape to print, the correct weight it should be, the customer’s address information, which carrier to use, and how to best get it to the customer on time – all with just a few scans and in a matter of seconds.
Fundamental to this operational dance are the warehouse’s battalion of smart robots. The most critical are arguably the Roomba-like delivery robots. Previously KIVA robots, Amazon bought out the company and renamed them Amazon Robots. Amazon competitors like GAP previously used KIVA robot technology and now no longer had access, giving Amazon a huge advantage.
These floor robots sort and deliver item stacks in the warehouses, using advanced calibration and weight sensors to know how fast to go, when to slow down, how to stay out of each other’s way, when to recharge, and how to read navigational QR codes on the floors.
Besides these short, round robots, there are large machines with heavy lift capacities, labelling robots, and more. Amazon’s sensors can tell when machines need service and quickly bring in needed repairs.
Though the Amazon robot army is in the thousands, human staff are still needed for picking and sorting items. Amazon has excelled in efficiency through very high picking standards and the gamification of their objectives, which incentivize the workers to hit targets and win prizes.
However, the stringent demand for high pick rates and accuracy comes at a price. In just 2019 alone, there were 14,000 injuries at Amazon fulfillment centers, which is double the industry standard. The only section of the business not overseen by AWS and cutting-edge tech might be where Amazon needs to improve next.