Boston Dynamics Wednesday announced a partnership Designed to provide improved strengthening learning to its humanoid robot from the electric atlas. The link is with Robotics & AI Institute (Rai Institute), previously known as Boston Dynamics Ai Institute.
The two organizations were founded by Marc Rabert, a former MIT teacher who was CEO of Boston Dynamics for 30 years. The Institute, founded in 2022, allows Raibert to continue the research which served as the basis for the dynamics of Boston.
The two have links with Hyundai. The Korean car manufacturer acquired Boston Dynamics in 2021; Hyundai also finances the institute, giving Raibert Free Rein to explore more experimental and bleeding technologies than possible in a commercial enterprise. The Institute reflects the creation by Toyota de Tri, or Toyota Research Institute, which announced its own partnership with Boston Dynamics in October, focused on the use of large models of behavior. (LBMS).
The twin partnerships are designed to improve the way in which the electric humanoid of Boston Dynamics learns new tasks. The agreement on robotics and AI of AI is specifically focused on learning strengthening, a method that works by trials and errors, similar to the way humans and animals learn. Reinforcement learning has traditionally been extremely exalted, although the creation of an effective simulation has made it possible to carry out many processes at the same time in a virtual framework.
The Boston Dynamics / Rai Institute Union began earlier this month in Massachusetts. This is the last in a certain number of collaborations between the pair, including a joint effort to develop a research research kit by strengthening for the quadruped spot robot of Boston Dynamics (which is its “dog” familiar robot ). The new work focuses both on the transfer of learning based on simulation in real world parameters and the improvement of the way in which the humanoid atlas of the company moves and interacts with physical environments.
Regarding the latter, Boston Dynamics points to “dynamic race and complete handling of heavy objects”. The two are examples of actions that require synchronization of the legs and arms. The biped factor of humanoid has a number 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.
The largest image, Note Raibert in a press release: “Our Rai objective is to develop a technology that allows future generations of smart machines. Working on Atlas with Boston Dynamics allows us to make progress in learning to strengthen the most sophisticated humanoid robot available. This work will play a crucial role in the progress of humanoid capacities not only by widening its skills, but also by rationalizing the process to acquire new skills. »»
The news of the partnership arrives one day after the figure that the founder and CEO of the AI, Brett Adcock, announced that the Bay Area Company abandoned a partnership with Openai in favor of the development of its own internal models.
“We have found that to resolve a large -scale embodied AI in the real world, you must integrate vertically Robot IA,” the executive told Techcrunch. “We cannot subcontract AI for the same reason that we cannot outsource our equipment.”
The figure has determined that the best models of AI for its humanoid are those which are developed specifically for its robots, internally. Openai’s approach to embodied intelligence – which means that AI in physical form – has naturally been less targeted given the size and scope of the Chatgpt Maker. This news also arrived on rumors that Openai explored the creation of its own humanoid robots.
Most of the companies involved in the humanoid space work on their own tailor -made AI models. This certainly applies to Boston Dynamics, which has decades of software development experience to operate on its own unique robotics systems. Although the RAI Institute is technically a distinct organization, both share a parent company, a founder and, probably, common objectives.
