Amazon says fully automated shipping warehouses are at least a decade away

The future of Amazon's logistics network will no doubt involve artificial intelligence and robotics, but it is an open question at what point AI-driven machines will do most of the work. According to Scott Anderson, director of robotics compliance at the company, the point at which an Amazon warehouse is fully automated end-to-end is at least 10 years old. Anderson's comments, published today by Reuters highlight the current pace of automation, even in environments conducive to robotic work, such as an Amazon warehouse.

In their current state, robots in the workforce are mainly competent in specific and repeatable tasks for which they are programmed with precision. Getting the robot to do something else requires costly and time-consuming reprogramming. And robots that can perform multiple different tasks and operate in dynamic environments that require the robot to see and understand their environment are still firmly in the field of research and experimental testing. Even the simple process of identifying an object and lifting it without ever having seen it before requires a series of complex and sophisticated software and hardware that does not yet exist commercially.

So, while a robot can help make a microchip and the body of a Tesla motor vehicle, it does not is capable of doing human tasks that warehouse work requires. At Amazon's facilities and compliance centers of other companies, a large part of the work is still largely in human hands, because it is difficult to train robots to see the world and use robotic pliers with the skill of human workers.

But as part of the ongoing deep learning revolution that has accelerated the progress of AI research over the past decade, robots are beginning to gain levels of vision and motor control that approximate human levels of sophistication . Amazon is one of the pioneer companies in this type of robots, and has the so-called annual collection challenge, after the warehouse's term of picking up an object to move it to another part of the logistics chain, promotes advances in the field.

Several companies and research laboratories have also been progressing on that front. UC Berkeley has a robotics lab that has made substantial advances in the field, and its new low-cost robot, a pair of humanoid arms controlled by a central system called Blue, can perform complex manual tasks such as folding a towel thanks to an AI Enhanced vision system. The OpenAI research laboratory has also been using an AI training technique known as reinforcement learning to teach a robotic hand more precise and elegant movements, the types of movement that would require a robot to replicate a human in a warehouse . Kindred, a new company based in San Francisco, manufactures a robotic arm called Kindred Sort that is implemented in warehouses for the retailer Gap that uses a combination of human piloting and automation to make a dynamic selection of products.


Blue is able to perform complex tasks such as folding a towel .
Image: UC Berkeley

According to Reuters Amazon has 110 warehouses in the US. UU., 45 classification centers, and approximately 50 delivery stations, all of which employ more than 125,000 full-time warehouse workers. But only a fraction of that work is done by robots. At this time, the robots are simply too inaccurate and clumsy and require too much training to be deployed on the floors of the factories outside of very narrow use cases.

For example, Amazon uses small Roomba-shaped robots, simply called "units," primarily to deliver large stacks of products to human workers, following established routes around the warehouse. "In the current form, technology is very limited. The technology is far from the fully automated workstation we would need, "Anderson told Reuters that he visited an Amazon warehouse in Baltimore today.

Please Note: This content is provided and hosted by a 3rd party server. Sometimes these servers may include advertisements. does not host or upload this material and is not responsible for the content.