Figure AI's Humanoid Robots Wrap an 11-Month BMW Pilot

Two Figure 02 humanoid robots completed an 11-month deployment at BMW Group's Spartanburg, South Carolina plant, working ten-hour shifts five days a week to support the production of more than 30,000 BMW X3 vehicles — logging roughly 1,250 operating hours and moving more than 90,000 sheet-metal components in the process. The successor, Figure 03, has now taken over at the same plant for more complex logistics and sequencing work.

What the Robots Actually Did

Figure 02's specific task was the precise removal and positioning of sheet-metal parts ahead of the welding process — work BMW itself describes as demanding in both speed and accuracy, and physically taxing over a full shift. This was not a demo or a showcase task; it was sustained, real production work integrated into an active automotive assembly line for nearly a year.

What Changes With Figure 03

• Tactile-sensor hands and palm cameras, adding fine manipulation feedback the previous generation lacked
• Wireless charging, removing a physical tethering constraint on shift duration and floor placement
• Speech-to-speech audio, enabling more direct verbal coordination with human co-workers on the floor
• A shift in task focus from Figure 02's welding-adjacent parts handling to logistics and sequencing applications

Why an 11-Month Real Deployment Is the Actual Story

Humanoid robotics has generated plenty of demo-stage coverage — this is different. An 11-month sustained deployment, with a documented handoff to a next-generation model at the same site, is evidence of a real production relationship, not a one-off showcase. For manufacturing and logistics operators evaluating physical AI, the more relevant question this raises is not whether humanoid robots can work a shift, but how a multi-generation deployment and hardware-refresh cycle actually gets planned and budgeted for.