Audi makes a quantum leap in AI for quality
Audi is creating thinking factories using AI driven tools to boost efficiency in quality control across its global manufacturing operations, led by developments at the Audi Production Lab
Audi is intensifying its focus on AI tools in production in what it describes as “a quantum leap for efficiency, quality and adaptability across its plants worldwide”.
The carmaker says it is following a clear AI and digitalisation roadmap and transforming production by developing “thinking factories” in which AI supports employees precisely where it creates the greatest value.
“On our way to intelligent production, we are creating a symbiosis of Audi’s decades of production expertise, our own innovative strength and the knowledge of strong partners,” says Henning Löser, head of the Audi Production Lab, which is focused on deploying emerging technologies in real production situations.
The Audi Production Lab is described as a crucial accelerator in the carmaker’s adoption of AI tools for turning ideas into series production. The lab, located in Gaimersheim near the Ingolstadt assembly plant, is manned by a team of 25 specialists who evaluate emerging technologies and prepare them for deployment in real factory environments. The purpose of the lab is to find and test innovations that reliably help optimise efficiency, ergonomics, flexibility and quality in Audi plants.
“We are currently piloting sequencing with robots in our logistics supermarket testing field,” says Löser. “With this project, we are taking the next step toward fully automated picking processes within our supply chain. The pilot will continue through the end of the year.”
Welding inspection
The roadmap for digitalising vehicle production has already enabled Audi to identify more than 100 AI use cases, many of which are already in project or series operation. One example is the use of AI robotics in weld spatter detection (WSD).
At Audi’s Neckarsulm site in Germany, the WSD system, developed with technology provider Siemens, uses AI to detect possible weld splatter on vehicle underbodies. The metal deposited by weld splatter risks such problems as under-vehicle cable breakage during production. The AI image processing directs blue lights to mark any detected weld splatters on the underbody, a job that was previously carried out by an assembly operative. That automation of laborious tasks is a key aspect of the AI strategy at Audi (and the wider VW Group).
“AI robotics is not about replacing people, it is about empowering them – taking over monotonous or physically demanding tasks so that our teams can focus on work that is more ergonomic, more creative and more value‑adding,” says Löser.
The AI-supported WSD system will shortly go into series production at six plants in Ingolstadt, according to Audi.
The WSD system builds on Audi’s use of AI to analyse 1.5m spot welds on vehicles at its Neckarsulm plant. Production staff were using ultrasound to manually monitor the quality of resistance spot welding processes based on random analysis. That system covered 5,000 spot welds per vehicle (across 300 per shift). By applying AI, employees were able to focus on possible anomalies, and the new approach enabled them to control quality more efficiently and in a more targeted way. That development has also been rolled out to other plants in the wider VW Group network.
Monitoring manufacturing
Audi is using AI-supported real-time process monitoring in several production areas to automatically detect anomalies in production and predict any costly disruption at an early stage. That includes in paint shop processes.
ProcessGuardAI is an AI monitoring tool developed in-house by Audi to drive optimisation into manufacturing using machine and sensor data. Described by Audi as a cross‑plant platform P‑Data Engine, itIt unites various system and equipment data from production at a uniform quality level. This allows Audi data scientists to develop and scale AI applications quickly and efficiently.
“With this framework, we create the foundation for consolidating decades of expert knowledge and system/process data from the entire Audi production network – and make it available to every employee for higher quality, improved efficiency and more stable processes,” says Audi.
The carmaker reports that the pilot phase for two use cases is currently ongoing at the Neckarsulm paint shop. One is for dosage optimisation in pretreatment and the other in anomaly detection in cathodic dip coating (CDC). Introduction into series production is planned for the second quarter of 2026, according to Audi, adding that early fault detection simplifies manual work steps and reduces follow-up costs.
In the next stages of development Audi will be using ProcessGuardAI to provide data‑based action recommendations for employees to manage production issues with an app that is supported by agentic AI. The tool will be central in predictive maintenance and quality assurance across various VW Group plants where it will be used to monitor all manufacturing processes, according to Audi.
Saving time with IRIS
Last year Audi also started testing a tool called IRIS (Intelligent Recognition and Inspection System), which uses cameras to check whether labels with technical data are correctly attached to the vehicle being assembled. The AI system evaluates whether the right label is correctly attached to the right part, has the right information, that it is in the right position and in the right language for the vehicle’s destination country.
Audi says this supports employees who continue to make spot checks, but the automated IRIS label check saves roughly one minute of production time per vehicle.
The carmaker says that more than 30 suitable use cases were identified for IRIS, some of which have already been successfully implemented in several plants.
Audi says that the advantages of AI-based image processing are visible on a number of levels. As seen in the time saved, the technology secures processes and optimises workflows through short control loops. It also improves quality assurance through objective, traceable and consistent inspection results, as well as fulfilling regulatory documentation requirements.
At the same time, it relieves employees by automating monotonous inspection tasks, while at the same time eliminating the associated inspection time.
Audi is now moving on to the implementation of IRIS Next with VW Group partners and reports the technology offering wider application, which can be implemented more quickly and with less cost in series production. IRIS Next is based on modern deep learning models that run in a centrally managed cloud architecture, says Audi. Industrial cameras in production capture image data, which is encrypted and sent to the cloud, where AI models analyse it. According to the company IRIS Next is not restricted to predefined inspections because its AI models enable targeted adaptation to the specific requirements of each use case. It is currently being applied at the Ingolstadt and Neckarsulm plants, where several thousand labels are inspected using AI every day.
“As a platform solution with a modular architecture, IRIS Next is highly flexible and can be scaled across a wide range of inspection tasks and production areas,” says the carmaker.
In 2026, IRIS is set to be deployed at ten VW Group locations, and the system is also being tested in use cases outside of assembly, such as battery production, logistics and the press shop. That is helped by the tool having successfully qualified as an inspection tool in accordance with the requirements laid down by the German Association of the Automotive Industry (VDA) in its guidance on quality management (VDA QMC 5 part 3 – Capability of optical sensors and image processing). The AI‑based image‑processing software can now perform fully automated visual inspection of product characteristics, which opens up its potential as a quality assurance tool.
Group-wide applications
Working with the Volkswagen brand, Audi has been responsible for the AI robotics strategy across the wider VW Group since 2024 and, in total, there are currently more than 20 AI-robotics projects underway across the group worldwide. All VW Group brands are now working together on AI robotics applications, with a dedicated core team to develops the strategy, and more than 150 employees exchanging ideas across brands in an open community.
That includes working with Swiss robotics start-up Mimic to uncover the possibilities of flexible, learning-based automation enabled by AI, especially in areas where traditional methods fall short. Recently, the project began testing a system trained to master intricate, multi-step assembly tasks, including door assembly on the Audi Q6 e-tron.
“Our end-to-end pixel-to-action model, running on our bi-manual platform, is capable of performing complex, dexterous and long-horizon insertion tasks,” said Mimic in a posting on LinkedIn. “We are excited about where this work is leading and the potential it opens up for deploying flexible, learning-based automation across a broader range of industrial applications where conventional automation reaches its limits.”
The group is also looking at the
use of AI to support humanoid robotics. In 2025, Audi and its Chinese partner
FAW carried out a pilot project using humanoid robots with the Beijing
Innovation Center of Humanoid Robotics and UBtech Robotics. Audi reports that
the project provided it with valuable insights into the underlying technology
and this year will launch “several high‑impact pilot projects at [its] German
production sites”.
“With AI robotics, we are
rethinking automation, especially where it was previously considered
impossible,” says Löser. “The goal is to give robots environmental and
contextual ‘awareness’ through AI to overcome today’s limits of automation.”
