Deliverable D7.3 / Effects on Safety

The Hi-Drive safety impact assessment presented in this report addresses the high-level research question: “What is the impact of automated driving (AD) and its enablers on safety?” This forms part of the overall Hi-Drive impact assessment, which also includes the impact areas of mobility, efficiency and environment, transport system, and socio-economic aspects. The goal of the safety impact area is to prospectively assess the safety impacts of AD and its technology enablers for higher market penetrations of AD (10%, 30%, 50%). It provides the direct safety impacts for various driving scenarios, qualitative results on potential indirect impacts, and the scaled-up safety effects at European level.
The impact assessment was performed for two automated driving function (ADF) configurations: the BADF (baseline ADF without the introduction of enablers) and the EADF (ADF with technology enablers incorporated). Various enablers were considered within the Hi-Drive project, such as different vehicle-to-everything (V2X) communication applications or optimised localisation techniques. In addition, two manual driving baselines were considered: current traffic and full penetration of mandatory advanced driver assistance systems (ADAS).
The primary operational design domains (ODDs) for the ADFs were urban roads and motorways, including extensions enabled by the technology enablers. Rural roads were outside the ODDs of both ADFs for automated operation; however, effects arising from ADAS utilising the sensor setup and computational resources of the ADFs were also analysed.
To analyse the direct effects within individual driving scenarios, in-depth accident data from GIDAS, TASC, and VOIESUR were used for driving scenario derivation using B-approaches (modified cases) and C-approaches (statistical distributions) for baseline generation. A total of 33 conflict driving scenario categories were simulated using different simulation tools and one common OEM-neutral ADF model calibrated with Hi-Drive operations data. The resulting injury severities were estimated using injury risk functions derived from accident data.
The results show that, especially for longitudinal traffic, mandatory ADAS can already reduce crash rates significantly compared to today’s traffic conditions. ADF configurations both with and without enablers demonstrated further crash reductions in longitudinal traffic, especially in more complex driving scenarios such as urban intersection conflict, where mandatory ADAS showed only minor effects. In addition to these positive effects of the ADFs, new types of conflicts might be triggered by minimal risk manoeuvres (MRMs), which were simulated too.
Combined with results from traffic simulations revealing changes in exposure to certain driving scenarios, these results served as the basis for scaling up to European level. This scaling-up was performed using data from the CARE database for the most recent fully available year (2023). For a penetration rate of 30% ADFs within the relevant ODDs, based on vehicle kilometres travelled, the ADF with enablers could avoid around 18–19% of motorway accidents across different severities compared to traffic today. This corresponds to a reduction of 16–17% compared to the full mandatory ADAS baseline. For urban environments, the reductions were even higher, being around 22% and 18%, respectively. The enhanced ADAS of automated vehicles outside ODD could achieve a reduction of 11–12% compared to traffic today and 5–7% relative to the mandatory ADAS baseline.
In addition, potential indirect safety impacts arising from the introduction of ADFs were investigated qualitatively based on a review of relevant literature, complementing the positive impacts investigated through simulations. Other indirect impact mechanisms may lead to both positive and negative consequences, such as a potential decline in driving skills or an increased amount of travel. The magnitudes of these indirect effects vary and are estimated to be small to medium in size.