Since unveiling its second-generation robot last August, Figure has made several significant steps towards its goal of a fully-autonomous household humanoid. Just weeks ago, the robotics company announced that was ending its partnership with OpenAI, claiming it had made major advancements with an in-house AI model. “Figure made a major breakthrough on fully end-to-end robot AI, built entirely in-house,” Founder Brett Adcock proclaimed on social media. “We’re excited to show you in the next 30 days something no one has ever seen on a humanoid.”
Today, I made the decision to leave our Collaboration Agreement with OpenAI
Figure made a major breakthrough on fully end-to-end robot AI, built entirely in-house
We’re excited to show you in the next 30 days something no one has ever seen on a humanoid
— Brett Adcock (@adcock_brett) February 4, 2025
Long before those 30 days were up, the company unveiled its breakthrough; a new AI model called Helix VLA (Vision-Language-Action). Figure says Helix is “the first Humanoid Vision-Language-Action model.” Having the ability to run on multiple robots simultaneously, Helix functions as sort of a “hive mind,” meaning all robots it is running on “learn” the same things together. In the Helix reveal video, Figure showcases two robots collaborating to sort and put away different items. The two robots work together in a way that is striking, and, more impressively, Figure claims that the two robots have never seen the items presented in the video before.
The following was originally published August 12, 2024:
It’s been an exciting several months for the humanoid robotics industry. From robots taking care of household chores and cleaning up the office, to new methods of “seeing,” it seems technology is trying to get helpful robots into our everyday lives at a faster pace each day. California-based company Figure, who recently made headlines by having its robots train to work at BMW manufacturing plants, have just revealed its second-generation robot, the F.02.
Like its first model, the F.02 utilizes a neural network to autonomously learn and complete tasks on its own. Information is fed to the neural network by an AI-assisted, multi-camera vision system. Figure says that the second-generation robot features a “ground-up hardware and software redesign,” affecting nearly all of the robot’s systems, including the AI system, sensors, cameras, and battery. Specifically, the company says the second-generation robot triples the computational power of its initial model. Interestingly, the F.02 features onboard microphones and speakers designed for one-to-one conversations with humans.
The F.02 is being aimed for commercial use, including production lines, but Figure says they are aiming to utilize their technology for an in-home assistant in the near future.
