Can the robot help you wash the dishes and sweep the floor as soon as you open your mouth? Figure launches the AI model 'Helix' targeting the household revolution.

Startup Figure AI has independently developed a new AI model ‘Helix’ for humanoid robots, which can control 2 robots simultaneously and handle unfamiliar items successfully. These features give Figure AI confidence to enter the blue ocean market of ‘home robots’. Founder and CEO Brett Adcock of humanoid robot startup Figure AI announced a new AI model ‘Helix’ for humanoid robots yesterday. Helix is a universal ‘Vision-Language-Action (VLA) model’ that uses visual and language commands to process information. Notably, Helix is the first dual-system VLA model that can control the upper body of humanoid robots rapidly and flexibly. This breakthrough allows robots to collaborate on long sequence tasks and handle unfamiliar items smoothly. Additionally, Figure demonstrated two breakthroughs in testing: successfully controlling groceries unseen during training and using the same Helix model weights for both robots without specific training or role assignment. The release of Helix signals Figure AI’s entry into the ‘home robot’ market, although training home robots may still require significant investment of money, manpower, and time. It is worth noting that Helix is still in its early stages, and the later cost expenditure remains difficult to estimate.