INJURY RISK-BASED CRITERIA FOR THE APPLICATION OF ADAPTIVE LOGIC TO ADAS SYSTEMS

Performance improvement of advanced driver assistance systems (ADAS) yields two major benefits: an increasingly fast progress towards autonomous driving and a simultaneous advance in vehicle safety. The high safety level provided by ADAS result primarily from the possibility to avoid possible impacts in correspondence of critical road scenarios. Nevertheless, specific obstacles (e.g., stationary vehicles, buildings) can interpose between the opponent vehicles and the working field of the sensors, weakening their functions: in these particular conditions, the impact can be inevitable (inevitable collision state – ICS). The systems currently available on the market are not capable to properly handle an ICS, because its occurrence is not conceived. In the present work intervention criteria for ADAS are introduced which are based on the vehicle occupants’ injury risk (IR), particularly useful in case of ICS. In a critical road scenario, the ADAS must first avoid the impact with maximum margin (maximum clearance between vehicles) and, in case of ICS, minimize impact severity and IR. Referring to a system capable of intervening on braking and steering, the ADAS must monitor the surrounding and act on the degrees of freedom adapting to the possible evolution of the scenario, following an adaptive logic. The sequence of optimal interventions based on such adaptive logic tends toward the best possible outcome. The context (model-in-the-loop) of the adaptive intervention employing the proposed criteria is first introduced, proposing a solution for testing its actual functioning (software-in-the-loop) with a view to its physical implementation (hardware-in-the-loop). The major criticality of the approach consists in the impact phase reconstruction, because IR is also a function of post-impact parameters (e.g., the velocity change ΔV experienced by the vehicle in the crash). To highlight the potential benefits offered by an adaptive ADAS and to monitor its behavior, a software has been developed based on the software-in-the-loop solution introduced. The best intervention selection is based on a database filled with results of simulations: the outcomes associated to each braking and steering intervention are summarized in the database, for many critical scenarios; the ADAS retrieves information from the database and, through IR-based criteria, selects the most favorable action. Testing the logic functioning in correspondence of three critical road scenarios in which two vehicles are involved, at each instant it is observed that the developed intervention logic aims at creating eccentrical impact configurations, associated to low ΔV; the low values of resulting impact severity demonstrate how the intervention criteria based on IR represent an important tool for the development of increasingly performing ADAS devices.