Robots/Parts/Training
Simulator, not named
A virtual world where a copy of the robot can practise. Figure says the part of Helix 02 that keeps the robot balanced learned everything in simulation, but it has not said which simulation software it used.
- 1 of 10
- robots in the database
- Not named
- made by
- Training
- brain layer
- Not public
- price
Key numbers
- Figure Helix 02
- the balance and whole-body movement layer, called System 0, was trained entirely in simulation
- Virtual worlds at once
- more than 200,000
- Software
- not named by Figure
- What it replaced
- 109,504 lines of hand-written code
Source: Figure, Introducing Helix 02: Full-Body Autonomy (figure.ai)
Strengths and trade-offs
Our analysisStrengths
Hundreds of thousands of practice runs can happen side by side.
A virtual robot can fall over endlessly without breaking anything.
Lets a learned skill replace a large amount of hand-written code.
Trade-offs
A simulator never matches the real world exactly, so real testing is still needed.
Because Figure has not named the software, outsiders cannot judge how it was done.
Needs a great deal of computing power.
Why robot makers choose it
Our readBalance has to be corrected many times a second, which is too fast and complicated to write out by hand. Practising in a huge number of virtual worlds at once lets a robot learn it in days rather than years.
This is our reasoning from the public facts. No maker has confirmed it.
Robots using it
This list covers only the robots researched so far. A robot missing here may still use this part.
Numbers and listed prices on this page come from the sources shown. Anything marked as our analysis, our read or our estimate is our own view and has not been confirmed by any company. Last checked 14 Oct 2026. How we verify