Abstract: A method of the collection of dangerous driving scene data and the evaluation of their complexity is proposed. Different from the collection method of natural driving data in large-scale real vehicle road test, the data before and after giving warning are collected as the data of dangerous driving condition for study based on advanced driving assistance system. On the basis of analyzing the dangerous driving scene data and their influence factors, the complexity evaluation indicator system of dangerous driving scenes is constructed in terms of environment, traffic flow and vehicle related factors. Based on the results of expert rating and applying analytical hierarchy process, the weights of indicators are determined and the model and method of evaluating the complexity of dangerous driving scenes are obtained and can be used for the simulation and test of intelligent vehicle

Key words: ADAS, dangerous driving conditions, scene analysis, AHP