Siemens and Nvidia Test Humanoid Robot in Live Factory
Siemens, Nvidia and Humanoid Ltd. tested the HMND 01 humanoid robot at Siemens’ electronics plant in Erlangen, Germany. According to Euronews and an Nvidia executive, the HMND 01 operated autonomously for more than eight hours, completed over 90% of its assigned tasks and moved about 60 containers per hour during the on‑site run.
The companies framed the trial as an exploration of physical AI for AI-driven factories, focusing on repetitive logistics work on the production line. Euronews noted the test was limited in scope to the Erlangen electronics plant.
Key takeaways (concise)
- Siemens and Nvidia tested HMND 01 at a Siemens Erlangen plant, running autonomously for more than 8 hours.
- Robot completed over 90% of assigned logistics tasks, moving about 60 containers per hour.
- Design and simulation-first development cut build time to about seven months from as long as two years; no timeline given for wider roll-out.
What happened in Erlangen: humanoid robot trial on Siemens’ factory floor
At Siemens’ electronics plant in Erlangen, Germany, a humanoid robot was trialed on the live shop floor to support routine internal logistics. The exercise took place inside the functioning electronics plant rather than in a laboratory setting, so the machine operated amid normal production activity.
The robot’s duties were limited to container handling and moving parts within the shop floor, carrying out straightforward logistics tasks alongside human workers. The trial focused on how such a device can assist existing staff with internal logistics, not on replacing human roles.
Siemens’ reported performance results
Siemens reported trial results from a live factory floor showing autonomous operation for more than eight hours, a task success rate above 90%, and throughput of around 60 containers per hour. These figures are specific to that live-floor trial and do not imply the same performance across other sites.
Who built what: Siemens integration, Nvidia AI stack and Humanoid’s HMND 01
According to company statements, Siemens provided the industrial integration backbone, Humanoid Ltd. handled deployment of the HMND 01 humanoid, and Nvidia supplied the AI and software stack that runs on the robot. That division of responsibilities left systems and factory integration to Siemens, while Humanoid focused on hardware and fielding the platform.
Nvidia framed its contribution under the “physical AI” concept, saying the stack spans simulation-first training through perception, reasoning and real-time edge inference, enabling robotics and edge AI workflows on the HMND 01, Deepu Talla, vice-president of robotics and edge AI at Nvidia, said. Siemens and Nvidia described the arrangement as pairing industrial integration with a simulation-to-edge AI pipeline to support safe, adaptive robot behavior alongside human workers.
Simulation-first development: reducing design time and physical testing
Companies are increasingly using a simulation-first approach to shorten development cycles and cut back on costly physical prototypes. Nvidia says its simulation and training tools allowed teams to perform virtual testing and robot training in a shared environment, letting engineers iterate designs faster and run scenarios that would otherwise require many hardware tests.
According to the company, those workflows helped bring design timelines down from as long as two years to about seven months, by moving much of the validation into simulation. More details on the platform are available at Nvidia Omniverse.
This figure is a company-stated estimate tied to the simulation-first approach and should not be read as an industry-wide benchmark, though simulation tools often help shorten parts of the development cycle and reduce the number of physical tests needed.
Why companies are testing humanoids: addressing labor gaps and tasks conventional automation struggles with
Companies say they are testing humanoid robots to address a labor shortage and to tackle tasks that conventional automation struggles with. Firms describe these machines as a form of flexible automation that can change roles on factory floors, in logistics and other industrial production settings, and as collaborators that can work alongside human workers.
Most deployments to date are framed by companies as trials rather than broad solutions, intended to assess real-world human-robot collaboration and whether flexible automation yields consistent gains. That distinction matters: stated business motivations are clear, while demonstrated outcomes remain limited to pilot projects and incremental results.
Worker training and safety protocols for humanoid robots
Trial-specific safety details were not disclosed, so public information is limited. In many deployments, clear human-robot interaction rules, sensor redundancy and fail-safes, an emergency stop and proximity limits, active supervision and incident reporting are required, alongside worker training and adherence to industrial safety standards such as those published by ISO.
What the companies said next, and what they did not
Siemens and Nvidia described the pilot as a milestone toward industrial reality, presenting the result as a step forward for industrial adoption. Both firms emphasized the significance of the trial while keeping public comments concise.
Neither company provided a rollout or deployment timeline; no schedule for wider roll-out was disclosed, and no commercialization dates were announced. Those limited disclosures are the only concrete next steps the companies publicly noted.
