Explainability & Determinism matters for successful ADAS/AD Development

Global automotive industry has reached some of the highest levels of quality in its history, as is evidenced by autonomous driving, electrification, connectivity, and other high-tech. The advancements of new emerging technologies have transformed the field significantly, in which Deep Learning plays quite an important role as enabler of many transformations. Nevertheless, there still exist many difficulties and limitations in implementing such technology in-vehicle system with respect to safety. At one hand, AI (deep learning-based) decision is ad-hoc and poses challenges for safety and hazard-free navigation. On the other hand, a complex software might lead to a non-deterministic system which would require a tremendous amount of engineering effort and huge dataset for confirmation and validation. In this talk, we will discuss the importance of explainability and determinism which matter for successful development of in-vehicle systems, and the balancing between high-tech and robust safety.