Typical pedestrian accident scenarios for the testing of autonomous emergency braking systems

Lenard, James; Danton, Russell; Avery, Matthew; Weekes, Alix; Zuby, David; Kühn, Matthias

Affiliations

¹Loughborough University, UK

²Thatcham, UK

³Insurance Institute for Highway Safety, USA

⁴German Insurers Accident Research, Germany

Abstract

The research objective of this work was to describe typical accident scenarios for pedestrian accidents.

The accident analysis forms a component of work by the AEB Test Group which aims to develop test procedures for assessing Autonomous Emergency Braking (AEB) systems. This technology is penetrating the vehicle market and is designed to offer protection against the occurrence and severity of collisions; however there is a need to evaluate the systems and their effectiveness since they are not yet subject to regulation or standardised assessment.

Case files for 175 pedestrians who were struck by the front of a passenger car were extracted from an in-depth accident database and reviewed in detail to establish the position and movement of road users before impact. A dataset of key parameters was formed from the detailed case reviews and subjected to a hierarchical cluster analysis to identify groups of similar accident scenarios. A second cluster analysis was performed on a dataset derived from the British national accident database for over 10,500 accidents where a pedestrian was struck by the front of a passenger car. This led to a second set of typical accident circumstances based on a comprehensive coverage of the accident population.

The national accident database for Great Britain, STATS 19, is compiled annually from police reports and effectively defines the national road accident population. In 2008 it registered over 28,000 pedestrian casualties from a total of around 230,000 road user casualties. The UK On-the-Spot (OTS) in-depth accident database was compiled by research teams at the scene of accidents in two regions of England from 2000 to 2010, including some non-injury accidents. Each team attended approximately 250 accidents per year, resulting in a total of over 4,700 accidents involving over 11,000 road users (including 288 pedestrians). This study was designed to collect a representative sample of accidents.

The cluster analyses show the association of accident circumstances such as speed limit, light conditions, weather, vehicle manoeuvre, pedestrian size, pedestrian movement, obstruction of line of sight, vehicle travel speed and change of speed to impact. The proportion of fatal, serious and slight casualties associated with these scenarios is quantified, showing for example that one scenario covered 12% of the population but 23% of fatal casualties.

Typical circumstances for pedestrian accidents in the dataset include (1) crossing from the kerb side without obstruction of the driver’s line of sight, (2) smaller pedestrians crossing from the kerb side with at least partial obstruction of the driver’s line of sight and (3) adult pedestrians crossing in inclement light and weather conditions. These scenarios were computed mathematically from large in-depth and national accident databases using cluster analysis and provide relevant information for the formulation of controlled tests of AEB systems.

Cited By (3)

Year	Entry
2013	Chauvel C, Page Y, Fildes B, Lahausse J. Automatic emergency braking for pedestrians effective target population and expected safety benefits. In: Proceedings of the 23rd International Technical Conference on the Enhanced Safety of Vehicles (ESV). May 27-30, 2013; Seoul, South Korea.
2013	Hulshof W, Knight I, Edwards A, Avery M, Grover C. Autonomous emergency braking test results. In: Proceedings of the 23rd International Technical Conference on the Enhanced Safety of Vehicles (ESV). May 27-30, 2013; Seoul, South Korea.
2014	Hamdane H, Serre T, Anderson R, Masson C, Yerpez J. Description of pedestrian crashes in accordance with characteristics of Active Safety Systems. In: Proceedings of the 2014 International IRCOBI Conference on the Biomechanics of Injury. September 10-12, 2014; Berlin, Germany.746-758.