What’s harder than building a humanoid robot that works on a factory floor? Building a factory floor that’s ready for one.
BMW just ran a humanoid robot – Figure 02 – on the production line at its Spartanburg plant for 10 months.
Ten-hour shifts, five days a week. It positioned sheet metal parts for welding across more than 90,000 components. Two robots worked in tandem – one operating while the other charged.
The transition from lab to production was, in BMW’s own words, faster than expected.
But here’s what none of the headlines about Figure 02 mentioned.
Before the robot touched a single part, BMW had to build something far less cinematic.
- Safety barriers.
- Wi-Fi 6 infrastructure.
- Standardised interfaces so the humanoid could plug into existing systems.
- IT, occupational safety, process management, and shop floor logistics.
All aligned from day one.
The robot showed up and did its job. But 12 months of unglamorous infrastructure work is what made that possible.
Now BMW is expanding to Europe – a pilot at its Leipzig plant with Hexagon’s AEON humanoid, targeting battery assembly and component manufacturing.
They’ve built a Centre of Competence for Physical AI in Munich to scale what they’ve learned.
And this is where it gets interesting.
AEON can learn a new task from just 20 demonstrations. But that only works because of what BMW already has – CAD models, digital twins, decades of documented production tasks.
“That’s not just institutional knowledge. That’s training data for physical AI.”
The factory floor isn’t just where robots go to work. It’s where they go to learn. BMW didn’t just prepare a floor. They’d been preparing a curriculum.
Sources: BMW Group press release (27 Feb 2026). BMWBlog on-site reporting, Plant Leipzig (March 2026).