Experimental approach to developing human driver models considering driver’s human factors

Abstract

A new approach to develop human driver models (HDMs) is proposed in accordance with the drivers’ generic human factors, i.e., gender, age, and experience, to develop more realistic vehicle simulations. The HDMs consist of three independent and stepwise models with functioning driver’s information processing stages based on the human factors: constructing drivers’ preview distance (PVD) models as a ‘cognition process’, implementing a finite preview optimal control algorithm as a ‘decision process’, and differentiating an ‘operation process’ according to neuromuscular efficiency. Eight different groups of 65 drivers with a 2 × 2 × 2 within-subject design participated in both the PVD estimates and neuromuscular efficiency tests to develop a set of statistically different HDMs. Regarding the preview distance models, an analysis of covariance (ANCOVA) procedure was adopted with two covariates (i.e., vehicle velocity and road curvature), while analyses of variance (ANOVAs) were performed on the neuromuscular efficiency parameters. The ANCOVA procedure produced eight significantly different cognition processes, whereas the ANOVAs revealed gender differences for the drivers’ neuromuscular systems. Moreover, an integrated vehicle simulation was configured with the HDMs using Carsim and Simulink software to observe the differential effects of both the cognition and operation processes on a double-lane-change (DLC) maneuver. During the simulations, gender differences in real-world DLC tests were also identified, especially between the male-oldexpert and the female-young-novice HDMs. The results presented in this study suggest that differentiating HDMs according to human factors is an essential process when utilizing vehicle simulations in the early stage of developing an intelligent vehicle system.