Boston Dynamics has announced a partnership with the Robotics & AI Institute (RAI Institute) to bring advanced reinforcement learning capabilities to its electric Atlas humanoid robot. The collaboration, which kicked off earlier this month in Massachusetts, aims to improve the way Atlas learns new tasks and interacts with its physical environment.

The RAI Institute, founded by Marc Raibert, a former MIT professor and Boston Dynamics' CEO for 30 years, is focused on exploring experimental and bleeding-edge technologies. The Institute is funded by Hyundai, which acquired Boston Dynamics in 2021. This partnership is the latest in a series of collaborations between Boston Dynamics and the RAI Institute, including a joint effort to develop a reinforcement learning research kit for the quadrupedal Spot robot.

Reinforcement learning, a method that operates through trial and error, similar to human and animal learning, has traditionally been extremely time-intensive. However, the creation of effective simulation has allowed many processes to be carried out at once in a virtual setting. The Boston Dynamics/RAI Institute partnership is specifically focused on transferring simulation-based learning to real-world settings and improving how Atlas moves through and interacts with physical environments.

Boston Dynamics highlights "dynamic running and full-body manipulation of heavy objects" as examples of actions that require synchronization of the legs and arms. The humanoid's bipedal form factor presents unique challenges and opportunities when compared with Spot. Every activity is also subject to a broad range of forces, including balance, force, resistance, and motion.

Marc Raibert notes that the goal of the RAI Institute is to develop technology that enables future generations of intelligent machines. Working on Atlas with Boston Dynamics enables the Institute 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.

The news of this partnership comes on the heels of Figure AI's announcement that it is abandoning its partnership with OpenAI in favor of developing its own in-house models. Figure AI's CEO, Brett Adcock, stated that to solve embodied AI at scale in the real world, vertically integrating robot AI is necessary. This shift highlights the importance of bespoke AI models tailored to specific robots, a strategy that Boston Dynamics has employed for decades.

The humanoid robotics space is increasingly focused on developing unique AI models that cater to individual robots. Boston Dynamics, with its extensive experience in developing software for its robotics systems, is well-positioned to benefit from this partnership with the RAI Institute. As the company continues to advance its Atlas humanoid robot, this collaboration is likely to have significant implications for the future of robotics and artificial intelligence.