Laundry-Folding Robots Are Finally Here -- and It's Backed by Jeff Bezos and OpenAI

By Alexandra Tremayne-Pengelly

Laundry-Folding Robots Are Finally Here -- and It's Backed by Jeff Bezos and OpenAI

San Francisco-based Physical Intelligence is working to tackle one of the most difficult tasks of A.I. application.

The advent of A.I. has already led to pivotal breakthroughs in the worlds of drug discovery, education and even chess. But incorporating the technology into everyday actions in the physical world remains a challenge. Physical Intelligence, a San Francisco-based startup, is working to solve this issue by creating general-purpose robots that can perform household chores ranging from folding laundry to clearing tables. Launched earlier this year, the startup is already valued at $2 billion after raising $400 million from a roster of investors including Jeff Bezos and OpenAI.

Sign Up For Our Daily Newsletter Sign Up

Thank you for signing up!

By clicking submit, you agree to our <a href="http://observermedia.com/terms">terms of service</a> and acknowledge we may use your information to send you emails, product samples, and promotions on this website and other properties. You can opt out anytime.

See all of our newsletters

The fundraising was first reported by The New York Times and confirmed by the company in a statement to Observer. The round was led by Bezos and venture capital firms Thrive Capital and Lux Capital. Other participants included Redpoint Ventures and Bond.

Instead of creating specialist robots programmed for a particular duty, Physical Intelligence is focused on building robots that can adapt to the environment for a variety of scenarios. "What we're doing is not just a brain for any particular robot," Karol Hausman, the startup's CEO and co-founder and a former Google DeepMind research scientist, told The New York Times. "It's a single generalist brain that can control any robot."

Robots that can fold laundry, bus tables and assemble boxes

https://observer.com/wp-content/uploads/sites/2/2024/11/PhysicalIntelligence_2min_v2_out.mp4

Physical Intelligence trains robots on models that combine the text comprehension and diverse data training of large language models (LLMs) with specific task datasets collected from robots. Last month, it unveiled a general-purpose robot foundation model called pi-zero that managed to successfully train robots to unload laundry, bus tables, place food in to-go containers and assemble cardboard boxes.

Folding laundry has been traditionally difficult for robots, given the complicated range of motions needed to put away a tangled pile of clothing. "To our knowledge, no prior robot system has been demonstrated to perform this task at this level of complexity," said the startup in a recent blog post.

Physical Intelligence's founding team include renowned engineers and scientists like Sergey Levine, a professor at the University of California, Berkeley, and Lachy Groom, a former executive at Stripe. In March, the company raised $70 million in a seed round that involved OpenAI, Thrive Capital, Lux Capital, Sequoia Capital, Outset Capital, Khosla Ventures and Greenoaks.

The company is currently hiring for ten roles including research scientists, machine learning engineers and robot operators. It also still has some kinks to work out, as evidenced by the startup pointing out instances when its robots failed at certain tasks like filling a bag with groceries or stacking serving bowls.

In its blog post, Physical Intelligence conceded that more work is needed in areas like autonomous self-improvement, robustness and safety. It hopes to improve over time with several collaborations across different companies and robotics labs that will redefine hardware designs and incorporate partner data to adapt models to specific platforms, said the company. Succeeding "will require not only new technologies and more data, but a collective effort involving the entire robotics community."

Previous articleNext article

POPULAR CATEGORY

entertainment

10215

discovery

4534

multipurpose

10586

athletics

10737