tools
Helix
Helix is Figure’s vision-language-action system for controlling humanoid robots in real time.
Helix combines perception, language understanding, movement, and learned control to operate Figure humanoid robots. It runs onboard and can perform tasks such as handling objects, navigating rooms, and coordinating with another robot.
The latest version described by Figure is Helix 2.5, which performs whole-body tasks such as tidying rooms, folding towels, and making beds across unfamiliar homes. Figure does not publish a Helix software price or offer it as a general-purpose API.
Features
- Processes perception, language, movement, and reasoning in one control loop
- Controls humanoid robots directly from visual input and language
- Performs whole-body locomotion and manipulation
- Supports dexterous grasping using tactile and visual sensing
- Coordinates long-horizon tasks without task-specific fine-tuning
- Generalizes tasks across unfamiliar homes and objects
- Supports multi-robot collaboration
Use cases
- Sort packages and orient shipping labels in logistics operations
- Unload and reload dishwashers across a kitchen
- Tidy living rooms and bedrooms
- Fold towels and make beds in unfamiliar homes
- Coordinate two humanoid robots on shared household tasks