Volkswagen Autoeuropa, Portugal

The jury awarded the "Digital Use Cases 2022" award to the Volkswagen Autoeuropa plant in Portugal, "The Home of T-Roc".

The Road Test Predictor is an application developed within the digital production platform of Volkswagen Autoeuropa. The aim is to reliably predict the outcome of the road test, which is achieved by using real-time production data along the in-house process chain from body shop onwards. It is the final quality control point before quality release and subsequent handover to the customer.

Up to now, 100% of the vehicles have undergone a final test before delivery to the sales network: the road test. When driving on different surfaces in combination with different driving situations, experienced employees check whether a vehicle shows fault patterns in its overall composition. These "live tests" increase the costs per vehicle significantly and have a negative impact on the CO₂ balance.

The primary objective of the project is to increase forecast reliability in such a way that the proportion of vehicles that must undergo this test can be reduced step by step without increasing the number of complaints in the field. As a secondary objective, knowledge should be gained to fully understand and optimise the processes.

The idea is simple and ingenious at the same time, but is a big step in using Big Data/Predictive Analysis to optimise processes. A large amount of data is generated in the production process of a vehicle. Can a pattern be discerned from this that allows prediction of the results of the road test in a few seconds? That was the challenge set by Volkswagen Autoeuropa employees in collaboration with partners such as HabberTec.

The application was enabled by several innovations within the Volkswagen Group, such as a scalable technical infrastructure or connectivity between the shop floor and the cloud.

The development of the Road-test Predictor took place in three steps: 1) understanding the actual data, 2) preparing the actual data for step 3, 3) modelling, evaluation, and selection of the forecast model.

160,000 vehicles with about 20,000 data from the production process were used as a basis (e.g., screwing torque and angle, measuring points). After an intensive validation phase, the Road Test Predictor went live for the first time in November 2021.

What does the process flow look like? After the last test step ZP7, the employee receives a decision from the Road Test Predictor within 40 seconds as to whether the road test should be performed. At present, this means that 10% of test drives no longer need to be carried out. This percentage will increase steadily as additional data sources are tapped because of continuous further development. The Autoeuropa plant has set itself the goal of increasing this proportion to 30%.

With the development of the Road Test Predictor, Volkswagen Autoeuropa is making an important contribution to the competitiveness of the entire Volkswagen Group, as the solution is provided as an APP via the "Digital Production Platform". Other plants are already in the pilot phase. The rollout will also provide the model with further data which will increase the forecast reliability. In addition, the project has created innovations for further digitalisation projects for the Group (e.g., water testing), which can be implemented in the lead of other brands or plants of the Volkswagen Group.

The jury sees the Road Test Predictor as an inspiring digital project for the entire automotive industry. The project offers high cost-cutting potential and an important contribution to reducing CO₂ emissions for the Volkswagen Group. The jury is looking forward to the further development of the Road Test Predictor from Volkswagen Autoeuropa.