Why Amazon Wants To Kill The Barcode: It’s possible that robots are the way of the future, however, it seems that robotic arms aren’t very adept at scanning barcodes. Barcodes can be difficult to locate and can be attached to objects with unusual shapes, which is something that robots are not particularly good at troubleshooting.
As a direct consequence of this, Amazon announced on Friday that the company is working on a strategy to do away with barcodes.
The e-commerce giant Amazon has developed a camera system that can monitor items as they move one-by-one down conveyor belts to ensure that they match their images. These pictures of items in Amazon warehouses were used to train a computer model, and the system was then used to create the camera system. Amazon’s artificial intelligence (AI) specialists and roboticists have the long-term goal of eventually combining the technology with robots that can detect objects while picking them up and rotating them.
According to Notes Antonakos, an applied science manager in Amazon’s computer vision group in Berlin, “Solving this challenge is crucial,” since it will allow robots to pick up products and process them without the need to locate and scan a barcode. It will make it easier for us to deliver packages to customers in a timely and accurate manner.
This method, which is known as multi-modal identification, is not likely to completely replace barcodes any time soon. Products stored in Amazon warehouses will be required to have barcodes so long as third-party manufacturers and shippers of those products continue to rely on the technology to identify and track their stock.
New Technology Already In Use In Some Countries
According to the company, their new technology is already being used in facilities in Barcelona, Spain, and Hamburg, Germany, and is speeding up the amount of time it takes to handle goods in those locations. Because this technology will be shared across all of Amazon’s businesses, it is feasible that one day you will see a version of it at Whole Foods or another Amazon-owned company that operates physical stores.
Amazon has integrated computer vision technology into a variety of its products. You can obtain assistance identifying items around the house by asking an Echo Show smart display, “Alexa, what am I holding?” to find out what you are carrying. Show and Tell is the name of the function that was developed for those who have problems with their vision in mind.
AI elements have also been included in cameras and photo apps by companies that specialize in smartphones and social media platforms. For instance, these apps can automatically classify photos.
According to Amazon, the issue of faulty items being shipped to customers, which was eliminated by the system, does not occur very frequently. When one considers the sheer number of products that are processed by a single warehouse in a single day, however, even seemingly little errors might result in enormous delays.
The artificial intelligence specialists at Amazon had to begin by compiling a library of photos of different products, which was something the corporation had never done before because there was no incentive to do so. The first versions of the algorithm were fed not only the photographs themselves but also data about the products’ size, and the cameras continue to acquire new images of items to train the model with.
When it was originally put into operation, the algorithm had an accuracy rate of between 75% and 80%, which Amazon regarded to be a promising beginning. According to the company, the accuracy is currently at 99%. The first problem the system encountered was that it was unable to distinguish between color variations.
During a promotion that took place on Prime Day, the system was unable to differentiate between the two distinct colors of Echo Dots. One of the dots on each of the packets was either blue or gray, and that was the sole distinction between them. The identification system required some retooling so that it could now assign confidence scores to its ratings so that it would only indicate items that it was extremely confident were erroneous.
The Artificial Intelligence (AI) team at Amazon has stated that it will be difficult to fine-tune the multi-modal identification system to evaluate things that are being handled by people. As a result, the ultimate goal is to have robots touch the products instead of people.