Qualification Challenges for Machine Learning Algorithms in Autonomous Driving Applications

Abstract

Within this talk I will motivate the usage of Machine Learning (ML) techniques for autonomous driving applications and simultaneously show current challenges within the automotive qualification processes. Although we see promising results of ML techniques for various tasks it is yet unclear how to qualify these algorithms for safety critical functions regarding the ISO 26262 standard. I will share our viewpoint on the qualification process and highlight some additional challenges which needs to be addressed before using ML techniques in safety critical applications.