---
source_type: official_company_writeup
title: "Introducing Helix 02: Full-Body Autonomy"
publisher: Figure AI
published: 2026-01-27
retrieved: 2026-05-17
url: https://www.figure.ai/news/helix-02
---

# Introducing Helix 02: Full-Body Autonomy

## Provenance

This raw source note records the official Figure AI technical writeup for Helix 02. The source is a company-published web article rather than an arXiv paper, so this file preserves the verified source URL, publication date, and extracted technical facts needed by the wiki source page.

## Extracted Technical Facts

- Helix 02 is presented as an extension of Helix from upper-body control to full-body humanoid loco-manipulation.
- The official writeup describes a System 2 / System 1 / System 0 hierarchy.
- System 2 reasons about goals, scenes, language, and behavior sequencing, then produces semantic latents for System 1.
- System 1 is described as a Transformer conditioned on System 2 latents that outputs full-body joint targets at 200 Hz.
- System 0 is described as a learned whole-body controller: a 10M-parameter neural network that takes full-body joint state and base motion as input and outputs joint-level actuator commands at 1 kHz.
- Observations include head cameras, palm cameras, fingertip tactile sensors, and full-body proprioception.
- Actions/control inputs cover complete joint-level control of legs, torso, head, arms, wrists, and individual fingers.
- Figure reports a 4-minute autonomous dishwasher task, 61 ordered loco-manipulation actions, bimanual transfers, stacking, and dexterous manipulation tasks such as unscrewing a cap, extracting a pill, dispensing syringe volume, and singulating small metal objects.

## Caveats

- The source is not a peer-reviewed paper and does not publish model weights, datasets, detailed benchmark protocols, ablations, or failure-rate statistics.
- The official writeup does not state that Helix 02 uses diffusion, flow matching, or a regression loss; those labels should not be inferred from this source.