Figure's Home-Data Strategy

Figure is trying to solve the consumer robot problem before it sells a consumer robot. The Brookfield partnership gives it access to video from 100,000 homes, which matters because home robots fail on edge cases like unfamiliar room layouts, clutter, and object placement, not on clean demo tasks. Combined with Helix and BMW factory deployment, Figure is building the training data, control software, and operating experience that could let it move into homes faster than peers starting with narrower datasets.

• The key asset is not just more footage, it is more variety. Figure said Project Go-Big uses Brookfield homes, offices, and logistics sites to capture goal directed human behavior at scale, and early results already trained Helix to navigate cluttered homes from natural language using only egocentric human video.
  1 figure 2 figure
• Industrial work funds the learning loop. BMW said Figure robots were piloted in Spartanburg production, handling sheet metal positioning across a ten month run that covered more than 30,000 vehicles. That gives Figure real uptime, safety, and teleoperation experience while consumer systems are still maturing.
  3 bmwgroup 4 figure
• The contrast with home first players is mostly about data collection method and price point. 1X is taking home orders with deliveries starting in 2026, while Sunday is using Skill Capture Gloves across 500 plus households ahead of a late 2026 beta. Figure is gathering household data earlier and at much larger environmental scale, even though its commercial robots are entering factories first.
  5 1x 6 sacra

The next phase is a race to turn household data into reliable low cost behavior. If Figure keeps compounding factory execution with home video pretraining, it can arrive at the consumer market with a robot that has already seen far more real homes than most rivals, which would shorten the path from industrial pilot to household product.