Boston Dynamics and Toyota Research Institute unveil AI-powered Atlas robot

Investing.com -- Boston Dynamics and Toyota (NYSE:TM) Research Institute (TRI) have demonstrated a significant advancement in robotics with their Atlas humanoid robot powered by a Large Behavior Model (LBM).

In a video released Wednesday, the Atlas robot performs complex tasks combining object manipulation with locomotion. The robot uses whole-body movements including walking, crouching, and lifting to complete packing, sorting, and organizing tasks.

The demonstration shows Atlas responding to unexpected challenges mid-task, such as when researchers close a box lid and slide it across the floor, requiring the robot to adjust its actions accordingly.

Unlike previous humanoid robots that separate walking and balancing control from arm manipulation, Atlas uses a single Large Behavior Model to control the entire robot, treating hands and feet almost identically.

This breakthrough stems from the joint research partnership between Boston Dynamics and TRI established in October 2024. The collaboration aims to accelerate the development of smart robots by combining the expertise of both organizations.

Scott Kuindersma, vice president of Robotics Research at Boston Dynamics, said, "This work provides a glimpse into how we’re thinking about building general-purpose robots that will transform how we live and work. Training a single neural network to perform many long-horizon manipulation tasks will lead to better generalization, and highly capable robots like Atlas present the fewest barriers to data collection for tasks requiring whole-body precision, dexterity, and strength."

Russ Tedrake, senior vice president of Large Behavior Models at Toyota Research Institute, highlighted the scalability of the approach: "One of the main value propositions of humanoids is that they can achieve a huge variety of tasks directly in existing environments, but the previous approaches to programming these tasks simply could not scale to meet this challenge. Large Behavior Models address this opportunity in a fundamentally new way – skills are added quickly via demonstrations from humans, and as the LBMs get stronger, they require less and less demonstrations to achieve more and more robust behaviors."

The project is co-led by Kuindersma and Tedrake, focusing on research to answer fundamental questions about humanoid robots and Large Behavior Models, advancing understanding of large models for whole-body control, including advanced manipulation and dynamic behaviors.

This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.