European factories are trialling humanoids. The honest reports from the floor are mixed and useful.

The gap between humanoid robot demonstration videos and factory floor reports remains wide. What has changed in the past year is that enough European pilots have run long enough to describe the gap precisely.

What the pilots report

On repetitive material handling in a structured cell, a humanoid is comfortably outperformed by a fixed arm that costs a fraction as much. No engineer interviewed disputed this.

On tasks that change weekly, in spaces built for people, with no time or budget to re-engineer the cell, the comparison changes. There the alternative is not a cheaper robot. It is no automation at all.

The three practical blockers

Uptime is the first. Pilots report meaningful time lost to recovery from situations a human would resolve in seconds, and the recovery usually requires a person anyway.

Safety certification is the second. A mobile machine of that mass working near people occupies an area of European machinery safety regulation that is still being interpreted, and plant safety officers are correctly conservative.

Cost per useful hour is the third, and it is the one moving fastest in the robots' favour.

Where the deployments are sticking

Not on the production line. In logistics adjacent work inside manufacturing sites: kitting, moving parts between cells, loading and unloading machines that were built for a human operator and are too expensive to replace.

That is a smaller market than the demonstrations imply and a real one.

The timeline engineers actually give

Asked when a humanoid would be a default purchase for a mid-size European manufacturer, the plant engineers interviewed gave answers clustered between five and ten years, and every one of them qualified it on safety certification rather than on capability.