Vision zero aspires to minimize fatalities or serious injuries due to human driver error, and it relies on high levels of automated driving, such as L3+. Higher levels of vehicle autonomy are achieved by combining several sensors via sensor fusion and other information based on connectivity from other vehicles or infrastructure. In this work, we focus on the former and in particular on monitoring the performance of ADAS sensors such as radars, lidars, and cameras. In the framework of this project, we will first develop and build the edge cloud infrastructure to support gathering of big data as well as the sensors´ digital twins. Then a robust communication between the cloud and the vehicle will be ensured, including elements of data anonymization and cybersecurity. The next step will be the core of the research project: the development of functions to assess the sensors perception quality as well as predict its performance degradation. Finally, we plan on testing the setup using a HiL validation platform made available by FEV Italia. This research project has the aim to develop functions for ADAS sensors perception quality assessment and performance prediction relying on AI techniques. These functions will run on an edge cloud infrastructure that communicates robustly with the vehicle.
