San Francisco-based AI-powered robotics startup Physical Intelligence announced this week it has raised $400 million in funding bringing its valuation to $2.8 billion.
Major investors include Amazon’s Jeff Bezos, OpenAI, Thrive Capital, Khosla Ventures, Lux Capital, Sequoia Capital and Bond Capital.
The startup aims to bring general-purpose AI into the physical world by developing foundation models and learning algorithms to power robots and other physically actuated devices, now and in the future.
Physical Intelligence has developed a general-purpose foundation model called π0 (pi-zero), which the company said is the first step toward its long-term goal of developing artificial physical intelligence. This would allow users to simply ask a robot to perform any task they want, just like they ask large language models and chatbot assistants.
“Like LLMs, our model is trained on broad and diverse data and can follow various text instructions,” Physical Intelligence stated in a blog post. “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.”
Those findings were released in a paper last week exploring the potential of robot learning to enhance flexible, general-purpose and dexterous robotic systems, aiming to address key challenges such as data, generalization and robustness.
The company is looking at a novel approach using “robot foundation models” to overcome these obstacles and tested its architecture by having robots do various tasks including folding laundry, cleaning tables and assembling boxes.
About the Author
You May Also Like