The rapid growth of Amazon's warehouse empire slowed this year as e-commerce demand cooled off. But that hasn't dissuaded the company from advancing automation efforts inside its facilities to improve operational efficiency.
Amazon unveiled several new warehouse technologies and provided updates to ongoing projects this year. Many are prototypes that have yet to be deployed at scale. Still, the intent is for them to make an impact in the company's logistics network one day.
“We don’t develop technology for technology’s sake,” said Siddhartha Srinivasa, director of Amazon Robotics AI, in an article on Amazon Science's website. “We want to develop technology with an end goal in mind of empowering our associates to perform their activities better and safer.”
Here is a rundown of six warehouse robotics innovations Amazon showcased this year as it looks to improve its fulfillment operations.
Cardinal is a prototype "robotic workcell" that selects one package out of a pile, lifts it and reads its label using artificial intelligence and computer vision. The system then places the package into a GoCart, non-automated wheeled transports used to move goods within facilities.
By using Cardinal to handle the lifting and turning of large or heavy packages, Amazon says the risk of employee injuries is reduced. A prototype handling packages up to 50 pounds is in testing, and the company expects the technology in fulfillment centers next year.
Stowing is a difficult task to automate, due to the dexterity required to handle a wide range of products in a delicate manner. Amazon has been developing a robotic system that can identify potential space in a storage bin and adjust its contents to create space before placing more items inside, per its Amazon Science website.
The company has prototypes of its robotic stow workstation at a lab in Seattle, Washington, and another system installed and handling live inventory at a Sumner, Washington, fulfillment center. This year, the system successfully stowed 94 of 95 items in a test featuring products with challenging attributes, like items with an offset center of gravity.
Sparrow is a robotic system that can detect, select and handle millions of individual products using computer vision and artificial intelligence. By moving items prior to packaging, it relieves employees of repetitive tasks and makes the fulfillment process more efficient, according to Amazon.
Sparrow is currently in the research and development phase, a spokesperson said in November. It's expected to complement other machines in Amazon's network — after items are handled by Sparrow and packaged, robotic arms like Robin and Cardinal can redirect the items within the warehouse.
Experimental transport robot
Amazon is testing free-roaming robots that handle the transport of oversized and unwieldy items within a fulfillment center. These items are normally transported by employees via pulleys and forklifts. The company aims to have the robots, which use artificial intelligence and computer vision, cover this task instead.
“This is the first instance of AI being used in autonomous mobility at Amazon,” Srinivasa said in the Amazon Science article.
As of October, dozens of these robots have been deployed in a few fulfillment centers for preliminary testing and refinement.
The fully autonomous Proteus, unveiled in June, lifts and moves GoCarts throughout fulfillment centers and sort centers, navigating around employees in the process. Amazon aims to automate GoCart handling to pare down the need for employees to manually move heavy objects.
Proteus, Amazon's first fully autonomous mobile robot, will initially be deployed in outbound GoCart handling areas, according to a company blog post.
The purpose of the prototype pinch-grasping robot is to move a variety of items quickly and without damage. Vacuum-like suction is the typical technology for robots moving a varied set of items, but there are challenges associated with suction. This includes a vacuum seal breaking prematurely or moving an item that requires contact on more than one surface.
In preliminary tests, the prototype achieved a tenfold reduction in damage on certain items without losing speed versus suction-using robots, according to Amazon Science's website. Amazon is still working to generalize the robot's manipulation capabilities to be viable for all items in its store, including for items weighing less than two pounds.