Validation Toolchain for Advanced Driver Distraction Warning Systems

European Union aims to significantly reduce the number of road accidents and casualties caused by driver's distraction, in this context, starting 2026, Advanced Driver Distraction Warning Systems (ADDWS) will be mandatory for new vehicles. Hence the demand for ADDWS validation frameworks has increased. The work done in this thesis aims to develop a proof of concept, based on state-of-the-art technologies, for a Validation framework that can offer variability and control over synthetic datasest, with focus on the quality of the animations. The dataset is meant to be used both in Hardware-In-The-Loop and Software-In-The-Loop fashion, in the second case requiring further studies over lens simulation, and is generated starting from a digital environment with controlled lighting, using some of the most advanced rendering features offered by Unreal Engine, and with digital avatars controlled through LiveLink and ARkit. A synthetic dataset however can not be used alone for validating ADDWS systems, since physical cameras may have an impact on the ability of the software to recognise a distracted occupant and precisely calculate important parameters such as eye gaze direction and head orientation. Taking into consideration such limitations, a proof of concept has been developed for Ameca Desktop, a SOTA robot capable of simulating most of the human expression, enabling hardware-in-the-loop simulations with synthetic subjects.