Amazon Warehouses Upgraded With Computer Vision Technology

Amazon Warehouses Upgraded With Computer Vision Technology

The automation of warehouse logistics is moving forward at pace – a reality further underlined by news that Amazon has introduced computer vision technology in its fulfilment centres. The online retail giant’s move towards full, or almost full, automation is a long held goal and has taken another step towards realisation.

Until now, fulfilment centre workers have scanned the barcode on each product that the premises, allowing that package to be tracked as they move through the warehouse. The process means robots know which large bins of assorted products to pick up and move to the next stage of the process, ahead of dispatch.

However, the days of hand-held scanners are over. One by one, Amazon’s fulfilment centres are AI-infused cameras and scanners that are able to pick up the barcode of each product as it is placed in its first bin. The new system then follows each product’s movements, cutting out the need for any human involvement.

At last month’s re:MARS artificial intelligence and robotics conference, Brad Porter, who holds the position of Amazon’s ‘vice president of robotics’, told the audience that the change introduced by the new scanners was small but significant in its impact on efficiency. He was at pains to stress that the new technology, which has now been rolled out across 20 of Amazon’s 175 international fulfilment centres, would not lead to redundancies among the 250,000 staff they employ on full time contracts. Rather it would simply make their job easier, meaning workers, or ‘associates’ as Amazon refers to them, would be able to handle bulky objects with two hands, without one being used to hold a scanner.

Amazon is introducing the latest technology in the world of computer vision elsewhere across its business. Examples include in delivery drones, internet-connected doorbells and an AI-powered fashion tool.

Computer vision technology allows software to not only record the outside world, like a camera does, but to draw conclusions from the visual data and take actions based on those conclusions. Practically speaking, its role is to ‘understand’ the images of the world outside that are being captured by cameras or scanners. Machine learning algorithms use huge amounts of data to learn the significance of different visual data points. Based on previous experience the algorithms can then recognise patterns in new visual data and make inferences.

The computer vision and supporting algorithms that allow automated scanning and tracking of products through an Amazon fulfilment centre are far from simple. They have to be able to pick up on subtleties such as a worker starting to put an object in a bin before deciding it is, in fact, too full, stopping the action mid-way and then putting it in another bin.

Amazon’s Amazon Web Services (AWS) cloud computing unit gives it a significant advantage when it comes to developing machine learning algorithms. The company is the world’s biggest provider of cloud computing as a service and is therefore able to take advantage of cost efficiencies when it comes to storing and processing the huge data volumes that effective machine learning algorithms need to ‘learn’.

Leave a Comment