Vehicle diagnosis currently requires extensive prior knowledge and a lot of time.
Development of a process
for automated detection of potentially defective components in a vehicle with the help of AI.
The complexity due to different vehicle types can be better managed and the performance of vehicle diagnostics thus becomes more efficient.
The company Hella Gutmann develops intelligent solutions for the repair and maintenance of cars and motorbikes of all makes and models, especially in cooperation with independent garages.
In the AI Marketplace, the company’s primary focus is AI-assisted vehicle diagnostics. Traditionally, potentially defective components in the vehicle are identified using fault codes and sensor values. Currently, a mechanic needs a lot of time in the workshop as well as extensive automotive knowledge in order to make a well-founded diagnosis on the basis of read-out error codes or measured sensor values (e.g. injection quantity). In addition, a large number of different vehicle brands are repaired in independent workshops, which increases the complexity.
Detecting defective components automatically with AI
The aim of this project is to develop and test methods for the automated detection of potentially defective components in the vehicle using artificial intelligence. In particular, suitable methods for data pre-processing and the training of machine learning models with the help of vehicle data
(fault codes, sensor readings and KM states) are to be developed and validated.
Furthermore, it is examined whether common models from the field of machine learning are suitable for application to vehicle diagnostics. Finally, these models are integrated into a prediction model for defective components. This prediction model will be optimised step by step through feedback from automotive experts and transferred into a demonstrator for AI-assisted vehicle diagnostics.
A demonstrator for the AI Marketplace
With this project, Hella Gutmann is developing a demonstrator in the form of a service that uses machine learning models to provide a list of potentially defective components for a specific automotive diagnostic case. In the future, the integration of this service into a product is planned. With the help of this AI solution, it will be possible in future for mechanics to better manage the complexity that arises from different vehicle types and thus also to carry out vehicle diagnostics significantly faster.