Audi is increasing its focus on AI tools in production, which it describes as “a breakthrough in efficiency, quality and adaptability across its factories around the world.”
The company says it is following a clear AI and digitalization roadmap and transforming production by developing a “thinking factory” that supports employees in the precise places where AI creates the most value.
AI-powered vision moves quality management from reactive to predictive
AI-driven inspection, edge computing, and digital twins are reimagining automotive manufacturing by predicting defects before they occur and enabling real-time process correction.
“On the path to intelligent production, we are creating a symbiosis of Audi’s decades of production expertise, our own innovative strengths and the knowledge of our powerful partners,” says Henning Racer, head of the Audi Production Lab, which focuses on the implementation of emerging technologies in real production situations.
The Audi Production Lab is said to be a key accelerator for automakers to implement AI tools to transform ideas into production. The institute, located in Geimersheim near the Ingolstadt assembly plant, is staffed by a team of 25 experts who evaluate emerging technologies and prepare them for implementation in a real factory environment. The aim of this lab is to find and test innovations that reliably help optimize efficiency, ergonomics, flexibility and quality at the Audi factory.
“We are currently piloting a robotic sequence in the test area of a logistics supermarket,” says Laser. “With this project, we are taking the next step towards fully automating the picking process within our supply chain. The pilot will continue until the end of the year.”
welding inspection
The roadmap for the digitalization of vehicle production has already enabled Audi to identify more than 100 AI use cases, many of which are already operational in projects or series. One example is the use of AI robots in weld spatter detection (WSD).
At Audi’s Neckarsulm plant in Germany, the WSD system, developed in conjunction with technology provider Siemens, uses AI to detect possible weld splatters on the underside of the vehicle. Metal deposits from welding splatter can cause problems during production, such as cable breaks under the vehicle. AI image processing shines a blue light to mark detected weld splatters on the underbody. Until now, this work was performed by assembly workers. Automating arduous tasks is a key aspect of Audi’s (and wider VW Group’s) AI strategy.
AI vision raises the bar for EV battery assembly
Atlas Copco’s VisionTools platform brings AI-enhanced machine vision to EV battery production, combining neural networks and traditional inspection logic to deliver unprecedented speed and quality assurance at scale.
“AI robotics is not about replacing people, but empowering them. It takes over monotonous or physically taxing tasks and allows teams to focus on more ergonomic, more creative, and more value-added work,” says Löser.
According to Audi, the AI-powered WSD system will soon be mass-produced at six plants in Ingolstadt.
The WSD system is built on Audi’s use of AI to analyze the 1.5 million spot welds on its vehicles at the Neckarsulm plant. Production staff used ultrasound to manually monitor the quality of the resistance spot welding process based on random analysis. The system covered 5,000 spot welds per vehicle (300 per shift). By applying AI, employees can focus on possible anomalies, and the new approach allows them to manage quality in a more efficient and targeted way. This development is also being rolled out to other plants within the VW Group’s wider network.
Production monitoring
Audi uses real-time process monitoring powered by AI in several production areas to automatically detect production anomalies and predict costly interruptions at an early stage. This also includes paint shop processes.
ProcessGuardAI is an AI monitoring tool developed in-house by Audi to drive manufacturing optimization using machine and sensor data. This engine, which Audi describes as the cross-plant platform P-Data Engine, integrates data from different systems and equipment from production at a uniform quality level. This will enable Audi’s data scientists to develop and scale AI applications quickly and efficiently.
“With this framework, we create a foundation that integrates decades of expertise and system/process data from across Audi’s production network, making it available to all employees to achieve higher quality, increased efficiency and more stable processes,” Audi said.
The car manufacturer reports that the pilot phase of two use cases is currently underway at its Neckarsulm paint shop. One for dose optimization in pre-processing and the other for anomaly detection in cathodic dip coating (CDC). Its introduction into series production is planned for the second quarter of 2026, Audi said, adding that early failure detection simplifies manual steps and reduces follow-up costs.
In the next stage of development, Audi will use ProcessGuardAI to provide database action recommendations for employees to manage production issues with an app supported by agent AI. According to Audi, the tool will be at the heart of predictive maintenance and quality assurance at various VW Group plants and will be used to monitor all manufacturing processes.
Save time with IRIS
Last year, Audi also began testing a tool called IRIS (Intelligent Recognition Inspection System). The tool uses a camera to check whether labels with technical data are correctly affixed to the vehicle being assembled. The AI system assesses whether the correct label is correctly attached to the correct part, contains the correct information, is in the correct position, and is written in the correct language for the vehicle’s destination country.
Audi says this supports employees who continue to perform spot checks, but the automated IRIS label check saves around one minute of production time per vehicle.
According to the automaker, more than 30 suitable use cases have been identified for IRIS, some of which have already been successfully implemented in several factories.
Audi says the benefits of AI-based image processing are clear on many levels. As seen in the time savings, this technology protects processes and optimizes workflow through short control loops. Objective, traceable and consistent test results also improve quality assurance and meet regulatory documentation requirements.
At the same time, by automating monotonous inspection tasks, we reduce the burden on employees and the associated inspection time.
How is quality control in automobile manufacturing changing?
In car factories, quality is no longer just inspected, but increasingly protected in real time. Cameras, AI, radar, lasers and digital inspection systems are intervening in production earlier than ever before. This transforms not only the final inspection but also the entire production logic.
Audi is currently working with partners from the VW Group to implement IRIS Next, reporting a technology that can be implemented faster and at lower cost in series production and offers a wider range of applications. According to Audi, IRIS Next is based on the latest deep learning models running in a centrally managed cloud architecture. Industrial cameras in production capture image data that is encrypted and sent to the cloud, where it is analyzed by AI models. According to the company, IRIS Next is not limited to predefined inspections, as its AI model allows for targeted adaptation to the specific requirements of each use case. It is currently being implemented at our factories in Ingolstadt and Neckarsulm, where thousands of labels are inspected using AI every day.
“As a platform solution with a modular architecture, IRIS Next is highly flexible and can be extended to a wide range of inspection tasks and production areas,” the automaker said.
In 2026, IRIS will be implemented at 10 VW Group locations, and the system is also being tested in use cases beyond assembly, such as battery production, logistics and stamping plants. This is achieved by the tool being certified as an inspection tool in accordance with the requirements laid down in the German Automotive Industry Association (VDA) Guidance on Quality Management (VDA QMC 5 Part 3 – Functions of Optical Sensors and Image Processing). AI-based image processing software can now perform fully automated visual inspection of product characteristics, expanding its potential as a quality assurance tool.
Group-wide applications
Audi, in cooperation with the Volkswagen brand, is responsible for the VW Group-wide AI robotics strategy from 2024, with a total of more than 20 AI robotics projects currently underway across the Group worldwide. All VW Group brands are now collaborating on AI robotics applications, with a dedicated core team developing strategy and more than 150 employees exchanging ideas across brands in an open community.
This includes working with Swiss robotics startup Mimic to uncover the potential for flexible learning-based automation enabled by AI, especially in areas where traditional methods are inadequate. Recently, the project began testing a system trained to master complex multi-step assembly tasks, such as assembling the doors of an Audi Q6 e-tron.
“Our end-to-end pixel-to-action model runs on a manual platform and is capable of performing complex, dexterous and long-lasting insertion tasks,” Mimic said in a post on LinkedIn. “We are excited about the direction this work is taking and opening up the possibility of deploying flexible learning-based automation to a wide range of industrial applications where traditional automation is reaching its limits.”
The group is also exploring the use of AI to support humanoid robotics. In 2025, Audi and its Chinese partner FAW collaborated with the Beijing Humanoid Robotics Innovation Center and UBtech Robotics to conduct a pilot project using humanoid robots. Audi reports that the project has yielded valuable insights into the underlying technology, and this year it “plans to launch several high-impact pilot projects.” [its] German production base.
“With AI robotics, we are rethinking automation, especially in areas that were previously thought impossible,” Laser says. “The goal is to give robots environmental and situational ‘awareness’ through AI to overcome the limitations of today’s automation.”
