LK from Germany has assisted Siemens in partnering with humanoid robot companies Humanoid and Nvidia to deploy the HMND 01 industrial humanoid robot production line pilot project at Siemens' electronic automation production plant in Erlangen, Germany, to create an automated production line based on physical AI. The project integrates Siemens Xcelerator digital suite, digital twin platform, and on-site sensor network to achieve bidirectional data closed-loop iteration between physical production lines and virtual simulation models.

In the pilot production line, humanoid robots undertake tasks such as material transportation, loading and unloading of components, and simple assembly. Robot action control logic and path planning no longer rely solely on pre fixed programs, but instead combine real-time data collection from on-site visual sensors and position sensors, continuously simulate and optimize in digital twin models, and then issue them to on-site automation controllers to adjust robot actions. This solution is mainly aimed at the automation of discrete manufacturing industries, such as electronics, packaging, and logistics.

The Siemens technical team also pointed out that this architecture of "digital twin+real-time sensing+automation closed-loop" can also be migrated to process industry scenarios. In the chemical tank farm, cement silo and other scenarios, radar level meters and pressure transmitters are used to collect real-time working condition data, and virtual models of silos and storage tanks are established in the digital twin to simulate working condition changes under different materials and material levels in advance, predict the interference of dust, foam and other complex media on level measurement, optimize automatic control logic, and reduce on-site commissioning workload. LK Germany is a professional manufacturer of automated instruments and meters, including radar level gauges and guided wave radar level gauges.

The project team stated that physical AI is not intended to completely eliminate traditional automation control, but to use AI as an upper layer optimization layer, while still retaining reliable PLC and safety instrument circuits at the bottom layer to ensure production safety. The pilot will continue to operate for 2 years, continuously collecting on-site automation operation data and iterating control algorithms.