Abstract

Road safety associated with vehicle operation is a complex function of dynamic interactions between the driver, vehicle, road and the environment. Using different motion perceptions, the driver performs as a controller to satisfy key guidance and control requirements of the vehicle system. Considerable efforts have been made to characterize cognitive behavior of the human drivers in the context of vehicle control. The vast majority of the reported studies on driver-vehicle interactions focus on automobile drivers with little or no considerations of the control limits of the human driver. The human driver's control performance is perhaps of greater concern for articulated vehicle combinations, which exhibit significantly lower stability limits. The directional dynamic analyses of such vehicles, however, have been limited either to open-loop steering and braking inputs or simplified path-following driver models. The primary motivations for this dissertation thus arise from the need to characterize human driving behavior coupled with articulated vehicles, and to identify essential human perceptions for developments in effective driver-assist systems and driver-adaptive designs.
In this dissertation research, a number of reported driver models employing widely different control strategies are reviewed and evaluated to identify the contributions of different control strategies as well as the most effective error prediction and compensation strategies for applications to heavy vehicles. A series of experiments was performed on a driving simulator to measure the steering and braking reaction times, and steering and control actions of the drivers with varying driving experience at different forward speeds. The measured data were analyzed and different regression models are proposed to describe driver’s steering response time, peak steer angle and peak steer rate as functions of driving experience and forward speed.
A two-stage preview driver model incorporating curved path geometry in addition to essential human driver cognitive elements such as path preview/prediction, error estimation, decision making and hand-arm dynamics, is proposed. The path preview of the model is realized using near and far preview points on the roadway to simultaneously maintain central lane position and vehicle orientation. The driver model is integrated to yaw-plane models of a single-unit vehicle and an articulated vehicle. The coupled driver-articulated vehicle model is studied to investigate the influences of variations in selected vehicle design parameters and driving speed on the path tracking performance and control characteristics of the human driver. The driver model parameters are subsequently identified through minimization of a composite cost function of path and orientation errors and target directional dynamic responses subject to limit constraints on the driver control characteristics. The significance of enhancing driver's perception of vehicle motion states on path tracking and control demands of the driver are then examined by involving different motion cues for the driver. The results suggest that the proposed model structure could serve as an effective tool to identify human control limits and to determine the most effective motion feedback cues that could yield improved directional dynamic performance and the control demands. The results are discussed so as to serve as guidance towards developments in DAS technologies for future commercial vehicles.