More Features, more Challenges: Rethinking Safety

As the number of vehicle features that are actively involved in vehicle control units continues to grow, the development of safety functions is becoming increasingly complex − and expensive. But AVL Software & Functions offers a solution:

AI, Data Science and Machine Learning

Neural networks enable significantly faster development than conventional approaches. They can also model highly non-linear functions that are typical for safety monitoring functions, e.g. with mappings and filtering techniques. Thanks to the flexible architecture of neural networks and machine learning, they can be optimally adapted to different safety requirements.

Our Approach

AVL has trained an annotated neural network to emulate the safety monitoring functions of a drive-by-wire system and predict the engine torque based on the driving requirements. The labeled training data used for this purpose was generated with an electric vehicle and tested for its safety suitability through detailed analysis.