01761600

Component

https://www-nrd.nhtsa.dot.gov/departments/esv/26th/

https://www-esv.nhtsa.dot...edings/26/26ESV-000113.pdf

One of the major challenges for enabling market introduction of automated driving is to identify risks and benefits of these functions. For this purpose, a new framework for assessing the safety impact of automated driving functions (ADFs) has been investigated. This framework is based on accident- and field operational test- (FOT-) data while using simulations for assessment of ADFs with respect to a certain baseline. According to the German Ethics Commission for Automated and Connected Driving, this baseline has to be human manual driver performance. For modelling of this baseline in simulations, so-called driver performance models are introduced in this publication and incorporated in an overall framework for effectiveness assessment. The main idea of the developed framework is that the types of driving scenarios, respectively physical accident constellations, do not change with automated driving. However, since ADFs are continuously controlling the behavior of the vehicle, it is possible that ADFs will get involved less frequently in accident scenarios playing a major role at human driving, e.g. rear-end accident scenarios. On the other hand, it is likely that other previously irrelevant accident types will rise. Consequently, the frequency of occurrence and the severity of the addressed driving scenarios may change with automated driving although the types of driving scenarios stay the same. To investigate the change of severity in a driving scenario, accident re-simulations are used. The changes in frequency of occurrence of driving scenarios are analyzed by using traffic simulations. In this work, so-called driver performance models are introduced for modelling human baseline in accident re-simulations. Key findings concerning the structure of these driver performance models are presented. The developed method and models are applied on two generic ADFs, a generic “Motorway-Chauffeur” (SAE level 3) and a generic “Urban Robot-Taxi” (SAE level 4). The results indicate that, e.g. a Motorway-Chauffeur at a market penetration of 50 % has a potential for reducing about 31 % of all accidents on German motorways resulting in personal injury. This equals 2 % of all accidents on German roads.

Paper Number: 19-0113-O

01760206

19-0113

English

11p

2019

26th International Technical Conference on the Enhanced Safety of Vehicles (ESV): Technology: Enabling a Safer Tomorrow

Digital/other

Figures; Photos; References; Tables

Drivers; Evaluation and assessment; Highway safety; Intelligent vehicles; Level 3 driving automation; Level 4 driving automation

Germany

Highways; Safety and Human Factors

TRIS, ATRI, USDOT

Dec 8 2020 4:02PM