Speaker(s): Giulia Padovan, David Brookes (Thatcham Research)
The research objective of the present work was to define the most frequent accident scenario use cases for rear-end impact and single vehicle scenarios and the most frequent contextual factors such as weather and light conditions for the use cases. The rationale for this analysis is that there is limited research on cluster analysis as a method for analysing road accident data.
The analysis will contribute to the specifications and development of test procedures for vehicles fitted with Advanced Driver Assistance Systems (ADAS), which are systems designed to support the driver in the driving process and hence increase car and road safety. The analysis considers the 2015 National accident database for Great Britain, STATS19, containing details about the accident, vehicles involved and casualties. Cluster analysis is an important tool when analysing traffic accident data, especially as a means of grouping similar uses cases, of which there could be many permutations, in a representative accident scenario.
Hierarchical cluster analysis has been performed in order to point out the most frequent accident causation and the most common scenarios involved in those kind of accidents. Moreover the insight provided by the analysis can be used to understand, at a more granular level, the impact on national accidents and motor insurance claims of new vehicle technologies being rolled into the national car parc.