Amazon launches new AI foundation model to power its robotic fleet, deploys millionth robot

Amazon Robotics announced the deployment of its millionth robot, building on its position as the world's largest manufacturer and operator of mobile robotics. This milestone robot was recently delivered to a fulfillment center in Japan, joining the Amazon global network that now spans more than 300 facilities worldwide.

The company also introduced a new generative AI foundation model. Amazon Robotics said it designed the model to make its entire fleet of robots smarter and more efficient.

Called DeepFleet, this AI technology will coordinate the movement of robots across the fulfillment network, improving the travel time of the robotic fleet by 10% and enabling the company to deliver packages to customers faster and at lower costs.

Making robots smarter

Built using the company’s extensive data sets of inventory movement within its sites and utilizing AWS tools, including Amazon SageMaker, the company this new AI model redefines fleet efficiency. It allows Amazon to store more products closer to customers, leading to faster delivery and lower costs. And, because it is built on AI that learns and improves over time, it will continue to find new ways to optimize how our robots work together.

“Think of DeepFleet as an intelligent traffic management system for a city filled with cars moving through congested streets,” said Scott Dresser, VP, Amazon Robotics, in a blog. “Just as a smart traffic system could reduce wait times and create better routes for drivers, DeepFleet coordinates our robots' movements to optimize how they navigate our fulfillment centers. This means less congestion, more efficient paths, and faster processing of customer orders.

10 years of robotic innovation

Amazon Robotics started in 2012 with a single type of robot that could move inventory shelves across warehouse floors. Today, we operate a diverse fleet of robots, designed to make its employees' jobs easier and safer, and its operations more efficient.

Amazon’s Hercules robots can lift and move up to 1,250 pounds of inventory. The Pegasus robots use precision conveyor belts to handle individual packages. And Proteus, the company’s first fully autonomous mobile robot, can safely navigate around employees in open and unrestricted areas of sites while moving heavy carts filled with customer orders.

“These robots work alongside our employees, handling heavy lifting and repetitive tasks while creating new opportunities for our front-line operators to develop technical skills,” Dresser said. “I'm particularly proud that since 2019, we've helped upskill more than 700,000 employees through various training initiatives, many focused on working with advanced technologies. In fact, at our next-generation fulfillment center, launched late last year in Shreveport, Louisiana, advanced robotics require 30% more employees in reliability, maintenance, and engineering roles.”

Technology to create real-world value

Amazon Robotics said DeepFleet represents its practical approach to AI innovation. Rather than pursuing technology for its own sake, the company said its focused on solving real problems.

For example, by reducing robot travel time by 10%, Amazon said it’s not just improving efficiency, but creating tangible benefits: faster delivery times, lower operational costs and reduced energy usage.

“The combination of our million-robot milestone and DeepFleet points to an exciting future where robotics and AI work together to reimagine what's possible in fulfillment and delivery,” Dresser added. “Twelve years ago, we worked backwards from a simple problem: how to help employees access inventory more efficiently. Today, we're applying the latest AI technology to make our entire robot fleet smarter, ultimately delivering better value to our customers through faster service and lower costs.”