What happened: Siemens trialled Humanoid’s HMND 01 Alpha at its Erlangen electronics factory, where the wheeled humanoid autonomously picked up, moved, and placed containers to support human operators.

Why it matters: The test hit practical metrics (60 container moves per hour, more than eight hours of continuous operation, and above 90% pick-and-place success), which is rarer than the usual “look, it walked” montage.

Wider context: The robot was engineered and trained using Nvidia’s robotics stack (Jetson Thor at the edge, Isaac Sim for simulation, Isaac Lab for reinforcement learning) and integrated into Siemens’ Xcelerator environment so it could coordinate with shop-floor workflows.

Background: The companies position it as an example of humanoids doing boring factory logistics, and say simulation-first development compressed the typical 18–24 month hardware cycle to about seven months.

Siemens trials Nvidia-powered humanoid robot for factory logistics — Verdict

Droid Brief Take: If humanoids are going to earn their keep, they will need to do tote-hauling and other unglamorous chores for hours at a time, not just backflip for investors. Siemens is at least measuring the right stuff, even if the hype still does the talking.

Key Takeaways:

• Throughput: Siemens and Humanoid reported the robot moved about 60 containers per hour during the logistics trial, a concrete operations number rather than a vibes-based “productivity” claim.
• Uptime: The HMND 01 Alpha ran for more than eight hours continuously, which matters because real factories do not pause to let your demo reset.
• Simulation Stack: Humanoid used Nvidia Jetson Thor, Isaac Sim, and Isaac Lab to design and train the system, and said simulation-first work cut development time from 18–24 months to around seven months.