If Physical Intelligence, an AI start-up seeking to create “brains” for a wide variety of robots, has its way, people will be able to ask robots to perform any task they want.
The company, which is joining the race to bring general-purpose AI into the physical world, announced November 4 that it had raised $400 million, in a financing round led by Jeff Bezos, Amazon’s executive chairman, and the venture capital firms Thrive Capital and Lux Capital. Other investors include OpenAI, Redpoint Ventures and Bond. The fund-raising valued the company, which was launched this year, at $2.4 billion, not including the new investment.
It is the latest example of how the prospects for robots that help with everything from household chores to handling hazardous waste have improved as progress in artificial intelligence accelerates and investment in the sector grows faster than anticipated. The total addressable market for humanoid robots is projected to reach $38 billion by 2035, up more than sixfold from a previous projection of $6 billion, according to a Goldman Sachs report.
Over the past eight months, Physical Intelligence has developed a general-purpose robot foundation model that it calls π0 (pi-zero). “We believe this is a first step toward our long-term goal of developing artificial physical intelligence, so that users can simply ask robots to perform any task they want, just like they can ask large language models (LLMs) and chatbot assistants,” the company said in a blog posting. “Like LLMs our model is trained on broad and diverse data and can follow various text instructions. Unlike LLMs, it spans images, text, and actions and acquires physical intelligence by training on embodied experience from robots, learning to directly output low-level motor commands via a novel architecture. It can control a variety of different robots and can either be prompted to carry out the desired task or fine-tuned to specialize it to challenging application scenarios.”
Building such a model requires a huge amount of data on how to operate in the real world. Those information sets largely do not exist, compelling the company to compile its own. Its work has been aided by big leaps in A.I. models that can interpret visual data.
Among the company’s co-founders are Karol Hausman, a former robotics scientist at Google; Sergey Levine, a professor at the University of California, Berkeley; and Lachy Groom, an investor and former executive at the payments giant Stripe.
In its blog posting Physical Intelligence explained the challenge and its vision: “The past decade witnessed practically useful AI assistants, AI systems that can generate photorealistic images and videos, and even models that can predict the structure of proteins,” it said in its blog. “ To paraphrase Moravec’s paradox, winning a game of chess or discovering a new drug represent “easy” problems for AI to solve, but folding a shirt or cleaning up a table requires solving some of the most difficult engineering problems ever conceived. To build AI systems that have the kind of physically situated versatility that people possess, we need a new approach — we need to make AI systems embodied so that they can acquire physical intelligence.”
Embedded in its blog posting are videos of robots folding laundry, clearing a table and more.
Industry investors say they are not surprised OpenAI backed Physical Intelligence. While most of OpenAI’s investments go to enterprise software players, some of its biggest investments have been in physical-world applications for AI, says market research firm CBInsights. OpenAI has invested in humanoid robotics developers Figure (which raised a $675 million Series B in February) and 1X (which raised a $23.5M Series A extension in March 2023). Both have partnered with OpenAI to integrate AI into their robots to enable reasoning, learning, and even language. Meanwhile, OpenAI just hired the former head of Meta’s AR glasses initiative Orion, Caitlin Kalinowski, to lead robotics and consumer hardware at OpenAI, with an initial focus on robotics.
Tesla, robotics company Boston Dynamics and Sanctuary AI are among others vying to build human-like intelligence in general-purpose robots. That’s not all. Earlier this year Nvidia, the U.S. maker of advanced AI chips, systems, and software, announced GROOT, a foundation model for humanoid robots.
Robots powered by GROOT (short for “Generalist Robot 00 Technology) will be designed to understand natural language and emulate movements by observing human actions—quickly learning coordination, dexterity, and other skills to navigate, adapt and interact with the real world, Nvidia said in its March 18 announcement. The new platform consists of a computer system that will power the robot and AI, plus a package of software including GenAI and other tools needed to build robots that are human-like, the company said at its annual developer conference.
“Building foundation models for general humanoid robots is one of the most exciting problems to solve in AI today,” Jensen Huang, founder and CEO of Nvidia said in a statement at the time. “The enabling technologies are coming together for leading roboticists around the world to take giant leaps towards artificial general robotics.”
IN OTHER NEWS THIS WEEK
FINANCIAL SERVICES
UBS Pilots Blockchain-Based Payment System
Finextra reports that UBS has successfully piloted a blockchain-based payment system, UBS Digital Cash, designed to improve the speed and efficiency of cross-border transactions for corporate and institutional clients. In the pilot, transactions with multinational clients and banks were successfully carried out, including domestic transactions within Switzerland and cross-border payments in US dollars, Swiss francs, Euros and Chinese yuan. In addition, the pilot also included the transfer of liquidity between various UBS companies. A key aim of the program is to improve the view of liquidity positions for corporate clients, which can be obscured due to delays in settlement timeframes. With UBS Digital Cash, the bank says companies should be able to manage intraday-liquidity and adjust liquidity buffers on their accounts more easily in the future, thanks to greater visibility of their total cash positions.
To access more of The Innovator’s News In Context stories click here
