Get the latest tech news
Helix: A vision-language-action model for generalist humanoid control
Figure was founded with the ambition to change the world.
Video 2: Helix allows for fast fine grained motor adjustments, necessary when reacting to a collaborative partner, while carrying out novel semantic goals.Helix's design offers several key advantages over existing approaches: When prompted to "Pick up the desert item", for instance, Helix not only recognizes that a toy cactus matches this abstract concept, but also selects the closest hand and executes the precise motor commands needed to grasp it securely. Helix displays strong object generalization, being able to pick up thousands of novel household items with varying shapes, sizes, colors, and material properties never encountered before in training, simply by asking in natural language.
Or read this on Hacker News
