Amazon has deployed its one millionth robot to a fulfillment center in Japan and is now rolling out a generative AI foundation model, called DeepFleet, to streamline how robots operate across its global logistics network.
DeepFleet is designed to coordinate robot movement across Amazon’s more than 300 global facilities, optimising traffic flow, reducing congestion, and cutting travel time by 10%. The model is trained on Amazon’s proprietary warehouse data using Amazon Web Services tools, including SageMaker.
“As DeepFleet learns from more data, it will continue to get smarter—driving deeper efficiencies, unlocking more selection closer to customers, and reimagining what’s possible in robotic logistics,” said Scott Dresser, VP of Amazon Robotics.
Amazon said the new AI model will contribute to faster and more cost-effective deliveries, further advancing the automation that has helped drive its record delivery speeds.
Amazon first deployed its own robots in 2021 and has since significantly expanded its robotics division, including:
-
Vulcan – a robot with a human-like sense of touch.
-
Digit – a bipedal, humanoid robot developed with Agility Robotics that mimics human movement and handling.
-
Unitree prototypes – humanoid robots reportedly being tested for doorstep delivery from Amazon Rivian vans.
The company also introduced seven new robots to assist with heavy lifting and repetitive tasks at a delivery station in Germany.
As Amazon inches closer to robot-worker parity in its warehouses, its automation ambitions appear far from slowing down.
Click here to sign up to Retail Gazette‘s free daily email newsletter
