PSI - Issue 24

Dario Vangi et al. / Procedia Structural Integrity 24 (2019) 423–436 D. Vangi et al. / Structural Integrity Procedia 00 (2019) 000–000

424

2

1. Introduction

The Vision Zero program of the European Commision (2011) actually represents an important reference for the advance of the automotive industry: the objective is the development of vehicles capable of assuring an increasing level of safety, allowing to reduce road fatalities to zero within 2050. The current design is based on the increase of vehicle’s passive safety - e.g., its crashworthiness - and, to a greater extent, in the performance of its active safety equipment. In fact, advanced driver assistance systems (ADAS) primarily allow reducing the probability for an impact between vehicles to occur - Spicer et al. (2018) - by functions as the autonomous emergency braking (AEB). In recent vehicles, the integration between several ADAS functions is frequently observed, aimed at providing a higher safety level as a consequence of the higher degree of automation achieved: in the case autonomous vehicles are considered, this complex ensemble of functions is identified as an automated driving system (ADS). The Society of Automotive Engineers (2018) (SAE) standard expresses the automation level of a vehicle, based on the tasks performed by the deployed ADAS functions. While SAE 4-5 prototypes exist, the circulating vehicles with the highest level of automation belong to SAE 3: in case of danger, the driver must be ready to take over control in place of the ADS, which however can manage the majority of the driving process (handover). The ADS can primarily insist on two degrees of freedom of the vehicle: • longitudinal acceleration - the vehicle traveling speed is modified. The most typical and commercially widespread system for controlling the vehicle longitudinal acceleration consists in the AEB, by which the vehicle decelerates employing the maximum available adherence (100%); moreover, modulation in the vehicle longitudinal acceleration can be achieved by adaptive cruise control (ACC); • transversal acceleration - the system intervention aims at modifying the vehicle degree of steering: systems as the lane keeping assist (LKA) or the more recent autonomous emergency steering (AES) respectively allow correcting the vehicle trajectory in case of deviation from the lane centre and prevent impacts with obstacles on the carriageway. The simultaneous action on longitudinal and transversal acceleration is mainly limited by the Kamm circle - Kaem pchen et al. (2009) - or analogously by the vehicle limits of longitudinal and transversal adherence (friction ellipse model described by Brach and Brach (2011)). Because of their higher ethical acceptability, the interventions aimed at decreasing the speed of the vehicle should be preferred to accelerations in high-risk road scenarios. Following the scheme proposed by Kullgren (2008), ADAS devices currently reduce road fatalities insisting on three leading factors: a) the number of collisions between vehicles, b) the impact severity and c) the injury level. In the first place, the number of collisions is reduced by the intervention on steering and braking to prevent the collision, by ADAS devices momentarily developed to intervene on a straight road; nevertheless, collisions in correspondence of intersections often occur, are much more complex and result in more critical scenarios in respect to straight roads. In these circumstances, the ADAS cannot prevent many accidents from occurring but can intervene to minimize the impact severity. An ADAS intervention aimed at reducing the closing velocity at collision V r is intuitively the best option for the impact severity minimization; however, V r only partially a ff ects the impact severity and the level of injury, which mainly depend on the acceleration experienced by the occupants - Gabauer and Gabler (2008). An alternative parameter to the occupants’ acceleration is the velocity change ∆ V sustained by the vehicle in the impact: ∆ V associated with real accidents can be more easily retrieved from in-depth accident databases and is significantly correlated to injury risk (IR) - Vangi et al. (2019a). The ADAS devices currently available on the market perform no evaluation on the severity of an impending impact; thence, they do not allow for road safety optimization acting on all the three factors cited above. Inevitable collision states (ICS) correspond to conditions in which no action by the ADAS or the driver can prevent the impact from occurring, as a consequence of insu ffi cient time to collision (TTC). Sundry reasons make ICS a reality in the current road environment. First, a circulating fleet with 11 years on average is observed on European roads by the European Automotive Manufacturers Association (2019), with co-existence of vehicles belonging to di ff erent SAE levels of automation: in such context, vehicle-to-vehicle communication will be limited also in the near future, and recogni tion of the opponents will chiefly depend on the sensors’ e ffi ciency. In contrast, obstacles like stationary vehicles or