AIMBOS

AIMBOS: Abstract Interpretation for Embedded AI Code Safety

Project leader: Dr. Peter Wägemann (FAU), Dr. Dominik Penk (Schaeffler), Dr. Dominik Riedelbauch (Schaeffler)
Project members: Tobias Häberlein (FAU)
Start date: 1. Januar 2025
End date: 31. Dezember 2027
Acronym: AIMBOS
Funding source: Schaeffler Technologies AG & Co. KG

Abstract

Artificial intelligence (AI), particularly in the form of machine learning models, steadily gains importance in industrial applications. For Schaeffler, the usage of those methods in edge- or embedded devices is of particular interest. Deployment and development of AI solutions in these environments presents a unique array of challenges. This project specifically focuses on proving functional code safety and the correctness of deployed applications. To this end, we will investigate how Abstract Interpretation, a well-known mathematical method to prove a wide range of program properties, can be used and extended for ML-based applications containing (deep) Neural Networks.