Figure unveils Helix, a vision-language-action model for household humanoids

Figure chief executive Brett Adcock revealed Helix, a generalist vision-language-action model that combines camera input with natural-language prompts to control a humanoid in real time. The company said Helix generalises well to unfamiliar household items and can coordinate two robots on a shared task, such as passing groceries between them in response to a voice command. Figure framed the home as a priority market, though TechCrunch noted that real domestic help is still likely years away. Helix illustrates how language models are becoming the conversational and reasoning layer of humanoid robots.

Source: TechCrunch