According to reports, Amazon is constantly developing advanced technologies based on artificial intelligence (AI). This technology can detect damaged goods in the packaging process, so as to remove them so that they will not reach customers.
It is understood that this process of detecting product status has been done manually in Amazon’s warehouse for many years. Up to five people will conduct a visual inspection at six o’clock to check the condition of the goods and whether they can be shipped. However, this is a time-consuming process for employees, who “rarely find damaged goods in inventory”.
To speed up the completion of this task, Amazon Fulfillment Technologies is developing a tool based on artificial intelligence, which can detect anomalies and mark defective products before they are delivered to customers. Last year, the research team found that they can use a machine learning model with reference images to teach the system how to compare the product it is looking at with the image of what the product should look like, so that the project can be started
In order to achieve this goal, scientists use computer vision technology to scan everything that passes through warehouses in the suburbs of the German capital. Then the AI model will analyze the scanning results to find possible damage. If any damage is not found, experts will analyze it and teach AI how to detect these errors.
Christoph Schwerdtfeger, applied science manager of Amazon Distribution Technology Company, said: “The efficiency of this system in identifying damaged products is three times that of manual identification.” We hope to deploy our damage detection software in 12 logistics centers in North America and Europe before Christmas. “Once implemented, this technology will help to detect the damage of more than 40 million products every month,” he said.
In addition to installing the system in more locations, the Amazon fulfillment technical team also plans to expand the functions of the system and carry out damage detection before shipping orders, for example, identifying the time and place where the damage occurred.
Post time: Aug-10-2023