Factory logistics bots
- Siemens, Humanoid and NVIDIA are working on wheeled factory robots for logistics tasks. - Early trials report navigation and pick‑place accuracy above 90% in controlled settings. - Vendors position these systems as replacements for repetitive floor transport and short hauls (x.com) (x.com).
Factories are starting to test wheeled humanoid robots for the short, repetitive jobs that move materials from one station to the next. Siemens said a Humanoid robot powered by NVIDIA software completed live logistics work at its electronics plant in Erlangen, Germany. (press.siemens.com) The machine, called HMND 01 Alpha, handled totes — the plastic bins used to move parts on factory floors — by picking, transporting and placing them for human workers. Siemens said the April 16, 2026 test met targets of 60 tote moves an hour, more than eight hours of uptime, and pick-and-place success above 90%. (press.siemens.com) This is not a general-purpose household robot. It is a factory vehicle with arms: a wheeled base carries it across the floor, and its manipulators do the handoff work that usually sits between conveyors, carts and human operators. (press.siemens.com) NVIDIA said the robot was developed with its Isaac Sim and Isaac Lab tools and runs Jetson Thor for on-robot computing. Those systems let companies train robots in software first, then move them into real plants after testing routes, grasping and task sequences in a digital model. (blogs.nvidia.com) Siemens has been building toward this in stages. In June 2025, it said its Operations Copilot would be extended to autonomous mobile robots and automated guided vehicles so engineers could assign transport jobs and configure routes through a software interface. (press.siemens.com) The Siemens-NVIDIA relationship also predates this trial. The companies expanded their manufacturing partnership on June 11, 2025, after first linking Siemens Xcelerator software with NVIDIA Omniverse in 2022 to build digital twins — virtual copies of factories used to test changes before they hit the floor. (press.siemens.com) NVIDIA has spent the past year pushing “physical AI,” its term for software that helps machines perceive, decide and act in real spaces rather than just analyze screens and text. At its March 16, 2026 robotics event, NVIDIA said companies including ABB, FANUC, KUKA and Yaskawa were also building on its robotics stack. (investor.nvidia.com) The immediate target is narrower than the hype around humanoids suggests. Siemens described the Erlangen deployment as logistics support inside an electronics factory, not full assembly-line replacement, and the published results came from a controlled industrial task with fixed totes and defined workflows. (press.siemens.com) NVIDIA is using Hannover Messe, which runs April 20-24, 2026 in Hannover, Germany, to show that kind of deployment as part of a larger manufacturing push. The pitch is that factories facing tighter labor pools and pressure for faster changeovers can use simulation, edge computing and mobile robots to automate the jobs between machines. (blogs.nvidia.com) For now, the clearest result is simple: one factory test showed a robot can spend a full shift moving bins with human-level consistency on a narrow logistics task. The next question is whether that performance holds outside a demo lane, across more plants, more layouts and more exceptions. (press.siemens.com)
