Boston Dynamics and the Robotics and AI Institute (RAI Institute) have announced a partnership to push the boundaries of humanoid robotics using reinforcement learning.
The collaboration will focus on enhancing the capabilities of the new electric Atlas robot, establishing a shared reinforcement learning training pipeline to develop dynamic and adaptable mobile manipulation skills.
Bringing together two of the leading research teams in robotics, the partnership builds on previous joint work, including the development of Spot’s Reinforcement Learning Researcher Kit. Released last year, the kit enabled unique locomotion training on the quadruped robot, leading to record running speeds of 11.5mph (5.2m/s). This latest project shifts the focus to humanoid robots, with several key objectives:
- Bridging the sim-to-reality gap: while simulation tools have become more advanced, transferring learned behaviours from simulations to physical robots remains a significant challenge. The teams will work on training policies that generate agile movements in real-world conditions, ensuring more robust and practical locomotion
- Enhancing whole-body manipulation: the ability to manipulate objects while moving – such as opening doors or operating levers – would greatly expand the usefulness of humanoid robots. The researchers will develop policies to improve stability and control in these tasks
- Developing full-body contact strategies: humanoid robots need to interact with their environment dynamically, whether running, lifting heavy objects, or performing complex movements. The project will explore reinforcement learning techniques to enable full-body coordination without requiring rigid control constraints
Robert Playter, CEO of Boston Dynamics, highlighted the significance of the work: “We are living in an extremely exciting time for humanoid robot development. But for humanoids to be useful, they must be flexible enough to work in many different kinds of environments and perform tasks in a wide variety of applications. Our collaboration with the RAI Institute brings together two of the world’s leading robotics organisations to help accelerate core capabilities needed to make robots like Atlas a valuable tool in people’s lives.”
Marc Raibert, Executive Director of the RAI Institute, echoed the potential impact: “Our aim at RAI is to develop technology that enables future generations of intelligent machines. Working on Atlas with Boston Dynamics enables us to make advances in reinforcement learning on arguably the most sophisticated humanoid robot available. This work will play a crucial role in advancing the capabilities of humanoids not only by expanding its skillset but also streamlining the process to achieve new skills.”
Boston Dynamics has long been at the forefront of robotic innovation, developing machines with extreme mobility and advanced manipulation skills. The electric Atlas represents years of integrated hardware and software development, providing a platform designed to push the limits of AI-driven robotics.
The RAI Institute, dedicated to solving key AI challenges in robotics, will contribute its expertise in perception, reasoning, and cognitive functions to enhance robot decision-making. This partnership will not only refine Atlas’s capabilities but also influence the future development of intelligent machines across various robotic platforms.
As the project progresses, Boston Dynamics and the RAI Institute plan to share updates, research findings, and demonstrations showcasing Atlas’s evolving abilities.
