Audi reportedly tests mobile humanoid at Ingolstadt plant

humanoid.guide reported Oct. 29, 2026, that Audi is testing parts sequencing with an Agile Robots wheeled humanoid at the Production Lab at its Ingolstadt plant. Audi announced a demonstration project at the plant on Oct. 21 with Agile Robots, Dassault Systèmes and NVIDIA. The project focuses on removing parts from containers and arranging them in the order needed for final assembly of Audi vehicles.

Audi says about 1 million picking operations take place each day, and this project targets the process of selecting parts from mixed containers and sequencing them for final assembly. The plan to use a virtual twin to assess safety and task duration could inform factory adoption decisions by evaluating not only the robot’s movements but also whether it can be integrated into existing logistics environments.

The project’s distinguishing feature is not simply bringing a humanoid into a factory, but targeting a specific logistics process—parts sequencing—and combining real-world testing with prior validation in a virtual environment. Audi’s figure of about 1 million picking operations a day indicates the scale of the process, but the announcement does not report results for processing speed or labor reductions. The demonstration at the Production Lab should be considered separately from deployment on a mass-production line. The target task involves picking parts from mixed containers, where parts may vary in shape, packaging and position. A mobile robot with two arms that works at human height is being adapted to an existing environment, testing whether it can handle variable conditions in current logistics spaces rather than merely replace fixed equipment. However, the announcement does not provide information on how the robot will work alongside employees, success rates by part, or stoppage frequency during continuous operation. Another feature of the process is the plan to build a virtual twin of the logistics environment before placing the physical robot, and to assess feasibility, task duration, ergonomics and safety. This adds a layer for evaluating the installation location and working conditions in advance, beyond planning the robot’s movements. Dassault Systèmes’ statement that LEO could reduce a process from several months to several days is a future vision from the company and should be distinguished from results achieved in this test. WorldOracle publishes benchmark comparing world-model predictions with real-world robot actions, published by this newspaper on Oct. 28, 2026, covered comparisons between model predictions and real-world execution. Audi’s project is not a benchmark; it is a plan to validate factory logistics conditions in a virtual environment, with a different evaluation target and purpose. A shared question is how well virtual assessments reflect actual operating conditions. The next points to watch are the range of parts included in the test, picking accuracy and processing time with the physical robot, recovery from faults, safety assessment results, and how validation in the virtual environment will be compared with real-world conditions. The announcement does not provide these results or specify the conditions for moving from the Production Lab test to a mass-production process. For now, it is best viewed as a demonstration project targeting a challenging process, not as evidence that automation has been achieved.

According to Audi, its plants conduct about 1 million picking operations a day to deliver the right parts to the right workstations in the right order. The mobile humanoid used in the test operates at human working height and has two arms. The project is designed to address recognition and grasping of parts in mixed containers and parts with varying positions and packaging, as well as the handling of semi-rigid parts. Before placing the robot in the factory, the project is creating a virtual twin of the logistics environment using Dassault Systèmes’ DELMIA. Feasibility, task duration, ergonomics and safety will be assessed on 3DEXPERIENCE, with libraries from NVIDIA Omniverse also in use. The announcement from Audi and its partners also outlined a future vision in which Dassault Systèmes’ AI-powered Virtual Companion, LEO, would support engineers and reduce processes that take several months to a few days.