Citation

Mandelík J, Bundzel M. Int. J. Crashworthiness 2019; 24(2): 221-234.

Copyright

(Copyright © 2019, Informa - Taylor and Francis Group)

DOI

10.1080/13588265.2018.1432740

PMID

unavailable

Abstract

We have performed experimental evaluation of the possibility to recognize the type of the vehicle vs. pedestrian collision based on the extent and localization of the pedestrian's injuries. The experiments verify the possibility to recognize the attributes of a collision to certain extent and the possibility to identify it based on these. We have used simulated data and a feedforward artificial neural network. The real world injuries can be parameterized by means of the FORTIS forensic system. We have simulated 255 collisions using the PC Crash program. We describe the process of data preparation , artificial neural network training and the experimental results that suggest that the individuality of pedestrian and vehicle collision parameters can be recognized.

Language: en

Keywords

Artificial Neural Network; identification of a collision type; injury parametrisation system; PC Crash; vehicle vs. pedestrian collision