Figure Helix VLA

Technical Analysis Helix demo video What is Helix? Helix is a generalist Vision-Language-Action (VLA) model developed by Figure AI for generalist humanoid control. Helix is designed to unify perception, language understanding, and learned control, overcoming major challenges in robotics, particularly the complexity of unstructured environments like a home. Key features and "firsts" of the Helix model include: Full-Upper-Body Control: It is the first VLA to output high-rate continuous control for the entire humanoid upper body, including wrists, torso, head, and individual fingers. Multi-Robot Collaboration: It's the first VLA to simultaneously operate on two robots, enabling them to work together on a shared, long-horizon manipulation task involving novel items. General Object Grasping ("Pick up anything"): Figure robots equipped with Helix can pick up virtually any small household object, including thousands they have never seen, simply by following natural language prompts. Single Neural Network: Unlike prior approaches, Helix uses a single set of neural network weights to learn all behaviors—from picking and placing to using drawers and cross-robot interaction—without any task-specific fine-tuning. Commercial-Ready: It is the first VLA to run entirely on onboard embedded low-power-consumption GPUs. The model represents a new scaling approach for humanoid robotics, aiming to translate the rich semantic knowledge captured in Vision Language Models (VLMs) directly into robot…