Helix

Helix, from Figure AI, is a Vision-Language-Action model for full upper-body humanoid control. Zero-shot generalization to new objects & tasks. Runs on embedded GPUs.

Helix, developed by Figure AI, is a groundbreaking Vision-Language-Action (VLA) model designed to bring humanoid robots to life with advanced language and vision capabilities. It enables full upper-body control, including wrists, torso, head, and fingers, and supports multi-robot collaboration for shared tasks. Helix stands out with its ability to generalize zero-shot to new objects and tasks, allowing robots to pick up virtually any household item using natural language prompts. Unlike traditional approaches, Helix uses a single neural network for diverse behaviors, eliminating the need for task-specific fine-tuning. It runs efficiently on embedded GPUs, making it ready for commercial deployment. Helix represents a significant leap in robotics, offering scalable, intelligent, and dexterous solutions for unstructured environments like homes.