Figure AI's bold move to part ways with OpenAI earlier this month has paid off. Today, the robotics pioneer unveiled Helix, a groundbreaking open source Vision-Language-Action (VLA) model that unlocks unprecedented capabilities for its humanoid robots.
Helix can be thought of as a robot brain that combines vision, language, and action to solve complex tasks. Helix enables multiple robots to work together on tasks they've never seen before and in the following demonstration Helix runs on two Figure robots at once, enabling seamless collaboration between the robots.
Helix runs on low-power GPUs locally on the robot hardware, making it ready for real-world applications.
Helix’s ability to enable AI to detect various objects quickly will enable Figure AI to make their robots useful in households, where objects are unpredictable.
Source: https://www.figure.ai/news/helix