Fuzzy Surrogate Safety Metrics for real-time assessment of rear-end collision risk. A study based on empirical observations

Konstantinos Mattas,Michail Makridis,George Botzoris,Akos Kriston,Fabrizio Minarini,Basil Papadopoulos,Fabrizio Re,Greger Rognelund,Biagio Ciuffo

DOI: https://doi.org/10.1016/j.aap.2020.105794

Abstract:The present paper discusses two fuzzy Surrogate Safety Metrics (SSMs) for rear-end collision, the Proactive Fuzzy SSM (PFS) and Critical Fuzzy SSM (CFS). The objective is to investigate their applicability for evaluating the real-time rear-end risk of collision of vehicles to support the operations of advanced driver assistance and automated vehicle functionalities (from driving assistance systems to fully automated vehicles). The proposed Fuzzy SSMs are evaluated and compared to other traditional metrics on the basis of empirical observations. To achieve this goal, an experimental campaign was organized in the AstaZero proving ground in Sweden. The campaign consisted of two main parts: a car-following experiment with five vehicles solely driven by Adaptive Cruise Control (ACC) systems and a safety critical experiment, testing the response of the Autonomous Emergency Braking (AEB) system to avoid collisions on a static target. The proposed PFS is compared with the safe distance defined by the well-known Responsibility Sensitive Safety (RSS) model, showing that it can produce meaningful results in assessing safety conditions also without the use of crisp safety thresholds (like in the case of RSS). The CFS outperformed the well-known Time-To-Collision (TTC) SSM in the a-priori identification of the cases, where the tested vehicles were not able to avoid the collision with the static target. Moreover, results show that CFS at the time of the first deceleration is correlated with the velocity of the vehicle at the time of collisions with the target.