Amazon Robots Are Efficient, But Struggles In "Target Picking"

Amazon's Autonomous robots are filling up their warehouses, where they are utilized to sort, load, and unload packages with impressive efficiency. Hwever, it sounds like they're still outmatched by humans when it comes to many essential tasks.

The e-commerce giant's robotic arm Sparrow, for example, excels at what's called "top-picking," or picking up an item at the top of a storage container. It can even manipulate over 200 million items of varying sizes and weights, Amazon claims.

But it struggles at "targeted picking," which involves having to search through a container to pluck out an item hidden by other stuff. It's a common task that any able human employee could do. For robots to do the same, however, will require nothing short of a breakthrough in the field.

"That's a really hard job," Tye Brady, chief technologist at Amazon Robotics, told The New York Times.

"I'm not saying it's impossible," he added, but that level of functionality would be "kind of the next frontier."

Even with their incipient capabilities, Amazon warehouses are already heavily relying on on robots, especially if their sheer quantity is any indication. The company has over 750,000 of the machines in use overall, per the NYT, or about half of its 1.55 million human workers.

In many cases, the robots do excel. A mobile robotic arm mounted on top of a wheeled platform called Stretch, created by Boston Dynamics, deftly unloads packages from the back of a truck and places them on a conveyor belt.

According to Sally Miller, global chief information officer at the shipping giant DHL, Stretch can unload around twice as many boxes per hour as humans, who might earn something like $17 an hour for the job. She did not say how much the robot cost, but gloated about its advantages over pesky human workers.

"It doesn't call in sick, and it can work for several hours," Miller told the NYT. "It's a great solution."

Meanwhile, Brady claimed that one of Amazon's new warehouses uses an automated inventory management system called Sequoia, boosting the speed of package processing by 25 percent compared to older ones, while being 25 percent cheaper.