Boston Dynamics Wednesday announced an association designed to bring better reinforcement learning to your Humanoid robot of the electric Atlas. The tie is with the Robotics & AI Institute (RAI Institute)previously known as the Boston Dynamics AI Institute.
Both organizations were founded by Marc Ribert, a former MIT professor who served as CEO of Boston Dynamics for 30 years. The Institute, founded in 2022, allows Ribert to continue the research that served as the basis for Boston dynamics.
Both have ties with Hyundai. The Korean automobile manufacturer acquired Boston Dynamics in 2021; Hyundai also finances the institute, giving Ribert Free Rein to explore more experimental and avant -garde technologies of what is possible in a commercial company. The institute reflects the creation of Toyota de Tri, or Toyota Research Institute, which announced its own association With Boston Dynamics in October, focused on the use of great behavior models. (LBMS).
Twin associations are designed to improve the way in which Boston Dynamics electric humanoid learns new tasks. The Robotics & AI Institute agreement focuses specifically on reinforcement learning, a method that works through proof and error, similar to the way in which humans and animals learn. Reinforcement learning has traditionally been extremely intensive over time, although the creation of an effective simulation has allowed many processes to be carried out at the same time in a virtual environment.
The Boston Dynamics/Rai Institute union began earlier this month in Massachusetts. It is the latest in a series of collaborations between the couple, including a joint effort to develop a reinforcement learning research kit for the robot of quadruped spots of Boston Dynamics (which is their family “dog” robot). The new work focuses so much on the learning transfer based on simulation to real world environments and improving how the company’s humanoid atlas moves and interacts with physical environments.
Belonging to the latter, Boston Dynamics points to the “dynamic career and full body manipulation of heavy objects.” Both are examples of actions that require synchronization of the legs and arms. The bipedal form of humanoid presents a series of unique challenges, and opportunities, compared to Spot. Each activity is also subject to a wide range of forces, including balance, strength, resistance and movement.
A general panorama, says Ribert in a statement, “our goal to RAI is to develop technology that allows future generations of smart machines. Working in Atlas with Boston Dynamics allows us to make advances in the learning of reinforcement on the most sophisticated humanoid robot available. This work will play a crucial role in the advancement of humanoid capabilities not only by expanding their set of skills, but also rationalizing the process to achieve new skills. “
The news of the association comes a day after the founder and CEO of Figure AI, Brett Adcock announced That the company of the Bay area is abandoning an association with OpenAi in favor of developing its own internal models.
“We discover that to solve the incarnate AI on the real world, you must integrate vertically robot AI,” the executive told TechCrunch. “We cannot outsourize the AI for the same reason why we cannot outsource our hardware.”
The figure determined that the best AI models for their humanoid are those that develop specifically for their robots, internally. The OpenAi approach for embodied intelligence, which means physical, has been less focused given the size and scope of the chatgpt manufacturer. That news also came after the rumors that Openai has been exploring the creation of its own humanoid robots.
Most of the companies involved in humanoid space are working on their own custom models. That certainly applies to Boston Dynamics, which has decades of experience in software development to run in its own unique robotics systems. While the RAI Institute is technically a separate organization, both share a parent company, founder and, presumably, common objectives.