Amazon Utilizes AI to Detect Faulty Products

by words Director of Applied The science of Amazon Robotics Jeremy Whayette, the company’s employees are now independently examining the goods for damage during their sorting and packaging. However, they often fail to notice defective products, since they are rare – less than 1 out of 1000 goods have damage, according to Amazon. On the other hand, Amazon processes billions of parcels per year, so the number of defective goods is quite large.

AI software works by scanning objects placed in boxes to detect signs of wear at special stations. If the goods are marked as damaged, the Amazon employee will check it again and decides whether it is possible to pack it and send it.

The technology was introduced in two warehouses. AI is 3 times more likely to detect damaged products than a person. Artificial intelligence was trained to determine the typical visual characteristics of defective goods, comparing the images of products without damage to those that had defects.

Amazon is interested in using more technologies for automating routine and monotonous work. Amazon previously introduced robots that could raise and move heavy objects, scan bar codes or place goods in carts. Cars cannot fulfill all roles and work with people.

Amazon believes that robots are the future, which is not surprising, given the fact that the company was criticized for a relatively high level of injuries in the workplace, complaints about the persecution of trade unions and violation of labor legislation. In addition, companies are difficult to hold workers.

/Reports, release notes, official announcements.