Weather and Road Geometry in Microscopic Acceleration Modeling: Calibration and Numerical Analysis Open Access
Downloadable ContentDownload PDF
AbstractThe increase in road traffic volumes leads to significant environmental and economic problems. It also gives rise to the growing demand for better traffic management strategies. Transportation researchers tried for decades to investigate the dynamics of traffic flow in order to optimize the movement of goods and people under different environmental conditions while reducing the environmental impacts and the economic losses due to congestion and traffic incidents. An important aspect missing from previous studies relates to the stochastic characteristics of the cognitive and psychological activities through which drivers process the dynamic information flux representing a given surrounding. Such activities include: perception, evaluation, judgment and execution. Moreover, most existing microscopic traffic models do not explicitly capture the interactions between individual drivers and external driving conditions. Recognizing these limitations in previous models, the objective of this thesis is to advance the state of knowledge in modeling longitudinal driving behavior while considering the surrounding environment's impact (weather and road geometry) on drivers' stochastic cognitive-based decision-making logic.In order to explore how a driver perceives the rapidly changing driving surrounding (i.e. different weather conditions and road geometry configurations) and executes acceleration maneuvers accordingly, this thesis adopts a prospect-theory (PT) based acceleration model that translates a subjective utility-maximization framework into longitudinal driving behavior. The PT logic in the acceleration model is extended by considering the different effects of the external driving conditions while keeping the probabilistic nature of the human perception-judgment-execution process. After an extensive effort reviewing the external factors that might influence drivers' decision-making, the major factors considered in this thesis include visibility level, road friction, curvature, gradient, median existence, lane width and shoulder width. By varying the values of those variables that parameterize external environmental characteristics, the probabilistic features of the longitudinal driving model with the corresponding acceleration probability density functions are presented. To better observe and study how drivers respond to different weather conditions and roadway geometries, driving experiments are carried out using a driving simulator-based approach. Foggy weather, icy and wet roadway surfaces, horizontal and vertical curves, different lane and shoulder widths, etc. are simulated while having participants drive behind a yellow cab at speeds/headways they feel appropriate. The produced experimental data include accelerations, speeds, longitudinal and lateral coordinates of the subject vehicle and the lead vehicle. Results indicate that, in general, drivers do not tend to follow the lead vehicle too far behind until visibility distance is drastically reduced by fog; however, the ability to follow a lead vehicle is maintained throughout most of the range of visibility distances. The distance gap between the subject vehicle and the lead vehicle decreases only when fog resulting in visibility distance of 65.62ft (minimum value adopted).After studying the driving trends observed in the different driving experiments, the extended PT acceleration model is calibrated using the produced trajectory data. The extended PT model parameters are able to reflect a change in the driving behavior when receiving different parameterized exogenous information. Two main parameters ψ and γ possibly reflecting external weather and road conditions - respectively - fluctuate considerably and do not change independently in each driving test. Some extreme values of these parameters ψ and γ are reached under "extreme" external conditions. To be more precise, two peak values of the hypothesized "weather parameter" ψ in dense fog condition of 65.62ft visibility distance and icy road surface are noticed. This shows a strong indication that in these extreme weather conditions; drivers put more weight on the weather impact and less weight on the road layout/geometry. Similarly, the hypothesized "road layout parameter" γ reaches its peak values in the cases of undivided road and narrow lane of 9ft. This result may indicate that drivers invest more attention and effort to deal with the roadway challenges compared to the effort to deal with the weather conditions. On the other hand, the calibration results suggest that drivers tend to underestimates the losses of a rear-end collision under adverse weather condition and overestimate the crash losses when traveling on the divided roads (metal barriers and Jersey barriers) and undivided road.