Simulation is one of the key enablers in the development of driver assistance functions, and going forward of autonomous driving. While there is an established effort within the automotive industry as a whole (and on a research level at the universities) to measure and quantify the sensor performance in the real world, the same rigour does not apply to the simulation platforms and their sensor modules. This talk highlights the challenges when quantifying simulation accuracy, proposes a universal approach, and details some numerical experiments performed in our working group.

Why quantify the simulation?
Three level of metrics
Lab-in-a-sim proposal
First results: MTF measurements