At this time of year, many of us have likely turned to Amazon to find gifts that Santa Claus or the Three Wise Men don’t have in their catalogs. And if there’s one thing we expect from the major logistics company, it’s efficiency and speed in shipping. Aware of how important these two things are to their customers, Amazon is thinking about how to streamline the entire process and be able to deliver orders in less time.
To achieve this, Amazon has considered eliminating barcodes from the products they ship and replacing this recognition system with one based on AI (Artificial Intelligence). This change addresses some drawbacks of barcode identification. Sometimes, for example, it’s not easy to find where they are, they can get damaged, or be difficult to read and scan if they are stuck to irregular surfaces.
The company is already working on the solution and testing it in its logistics centers in Barcelona and Hamburg: doing away with barcodes and even eliminating the manual identification process of items. To solve this “uncomfortable and inefficient” dependence on barcodes, as the company describes it, Amazon is betting on multimodal identification, known as MMID.
This new system consists of identifying products based on their appearance and dimensions. To achieve this, the first thing they had to do at the Hamburg and Barcelona centers was to take photographs of all the objects passing through the packing centers. With these photos, they built an image database so that not only the robots used by the company knew the product dimensions, but also to feed a machine learning system and an algorithm.
In the initial tests, Amazon observed that the new system was able to identify products with an accuracy approaching 80%, a very positive figure for an early stage. Still, the company states that this technology is still being improved and will take some time before it can be used. One of the main drawbacks is that currently, this system requires good lighting and the ideal speed on the conveyor belt so the cameras can take the necessary photos for the AI to do its job.
Amazon also points out another possible drawback: handling by an employee. This would also make object detection more difficult depending on how it is held. To address this, the company is also working on robotic arms to stop relying on humans for this specific task.
It seems this distribution giant is taking it very seriously. Now we just have to wait for their dream to come true and for us, the customers, to be the main beneficiaries.
