Analysis of the Correlation between Driver's Visual Features and Driver Intention 2019-01-1229
Driver intention recognition provides theoretical support for pre-judgment and collision avoidance maneuvers of dangerous traffic scenarios, which can effectively improve traffic safety. Research on the correlation between driver's visual characteristics and driving intentions can provide theoretical guidance for features selection of driving intention recognition. Natural driving tests on open roads and semi-closed roads were carried out. Scene video, driver movement status, and steering wheel angle signals are captured. A data set consisting of 117 straight-through segments, 179 left change segments, and 174 right change segments was created. Line-of-sight direction and head pose estimation algorithm based on machine vision is developed. The visual features of the driver's head and face are systematically extracted from the dataset, and the correlation between each feature and driving intention is obtained using statistical methods.