The next generation humanoid robot from Boston Dynamics will have Google’s DeepMind DNA

Robotics company Boston Dynamics has partnered with Google’s Artificial Intelligence Research Lab to accelerate development of the next generation robot Atlas — and make it behave more humane around people.

The partnership, announced Monday during Hyundai’s CES 2026 press conference, focuses on robotics research that will use core AI models from Google DeepMind. Boston Dynamics’ humanoid robot Atlas will be the first test case, according to Carolina Parada, senior director of robotics at Google DeepMind.

“We look forward to integrating our cutting-edge AI models with Boston Dynamics’ new Atlas robots, and will aim to develop the world’s most advanced robot foundation model to deliver on the promise of real human needs for general purposes,” Parada said on stage.

The tie-up comes less than a year after Google’s AI research lab announced new models of artificial intelligence called Gemini Robotics that are designed to allow robots to perceive, think, use tools, and interact with humans. Gemini Robotics is based on the large-scale multimodal generative AI model, Gemini. At the time, Google DeepMind said its robotics AI model had been trained to generalize behavior across a range of different robotics devices.

Enter Boston Dynamics, and majority owner, Hyundai Motor Group. While accelerating research will be a key part of this partnership, there is an intention to expand into the real world.

Boston Dynamics already has products, like the four-legged Spot, in the hands of customers in more than 40 countries. The Stretch warehouse robot has unloaded more than 20 million boxes globally since its launch in 2023, according to Hyundai. Now Boston Dynamics and Hyundai are gearing up for the next generation, starting with the humanoid robot Atlas, which the company announced Monday is already in production and headed to Hyundai’s factory in Savannah, Georgia.

A prototype of Atlas walked on stage during the press conference, showing off its mobility. But as Alberto Rodriguez, director of Atlas behavior at Boston Dynamics, points out, “turning Atlas into a product requires more than just athletic performance for humans to deliver on its promise. They have to be able to interact with people naturally.”

Rodriguez and his counterparts at Boston Dynamics believe that recent advances in artificial intelligence have created a clear path to achieving those capabilities. This type of natural interaction with humans has real safety implications.

The Atlas product, which was also unveiled on stage Monday and will eventually head to the Hyundai factory, has 56 degrees of freedom with rotating joints and human-sized hands that have tactile sensing. This is powerful. The Atlas robot can lift up to 110 pounds and is designed to perform repetitive movements.

With this kind of dexterity and power, it will be crucial for Atlas, or any humanoid robot, to interact and work safely with humans. Some of that has been handled on the hardware side; For example, the Atlas system has 360-degree cameras to allow it to know when people are approaching. But DeepMind’s work could help robots learn how to behave.

“Instead of having a set of pre-defined tasks loaded onto a robot, we believe robots should understand the physical world in the same way that we do,” Parada said. “They have to be able to learn from their experience. They have to be able to generalize to new situations and improve over time. So, whether it’s assembling a new car part or tying shoelaces, robots have to learn in the same way we do through a set of examples, and then improve very quickly with a little practice.”

Hyundai, which plans to bring the Atlas system to its plant this year and eventually deploy it for tasks such as parts sequencing by 2028, has developed protocols to increase safety and efficiency.

Hyundai said Monday it is opening a U.S. facility this year called the Robot Metaplant Application Center, or RMAC, that will teach robots how to map movements such as lifts and turns. Training data from RMAC will be combined with real-world data collected via a software platform used at its Georgia factory to continuously improve the robots.

Follow all of TechCrunch’s coverage of the annual CES conference here.

This article has been updated to include more information about Atlas specifications.