Modelling heavy vehicle lane changing
2017-02-08T04:10:59Z (GMT) by
Lane changing manoeuvres have a substantial impact on microscopic and macroscopic traffic flow characteristics due to the interference effect they have on surrounding vehicles. The interference effects of heavy vehicles’ lane changing manoeuvres on surrounding traffic are likely to be greater than when passenger cars execute lane changing manoeuvre. While heavy vehicles account for a minority of traffic stream, heavy vehicles have a pronounced effect on traffic flow and produce a disproportionate effect particularly during heavy traffic conditions. Heavy vehicles impose physical and psychological effects on surrounding traffic which are the results of physical and operational characteristics of heavy vehicles. The number of heavy vehicles on urban freeways has increased over the past three decades and this trend is likely to continue over the next decade. Despite the increasing number of heavy vehicles on freeways, previous studies have predominantly focused on the behaviour of passenger car drivers. In the previous lane changing models, heavy vehicles are accommodated in current lane changing models by calibrating the parameters of a general lane changing model for heavy vehicles rather than by incorporating a lane changing model developed specifically for the heavy vehicle drivers. However, heavy vehicle and passenger car drivers may have fundamentally different lane changing behaviour. In this study, the trajectory dataset is based on the video images of two freeway sections. In general, extracting the trajectory dataset from video images makes it impossible to capture some physical (e.g. weight) and operational (e.g. power) characteristics of vehicles. The length of vehicles is one of their physical characteristics that can be extracted from video images. Therefore, vehicle length is used to identify heavy vehicles in this research. The vehicles with the length of equal to or greater than 6 meters are classified as heavy vehicles. This classification is consistent with the definition of the heavy vehicles in the trajectory dataset used for this study. In this research, the lane changing behaviour of a driver has been characterized as a sequence of three stages including motivation to change lanes, selection of the target lane and the execution of the lane change. This research has provided new insight into the role that traffic parameters associated with the surrounding vehicles plays in the lane changing behaviour of heavy vehicle and passenger car drivers. From detailed examination of vehicle trajectory data, differences were identified in the lane changing of heavy vehicle and passenger car drivers in terms of the three stages of lane changing behaviour. To understand the influencing factors on heavy vehicle drivers’ lane changing, it is required to analyse the surrounding traffic characteristics at the time that the heavy vehicle drives change lanes as well as when they do not wish to execute lane changing manoeuvre. From detailed examination of the surrounding traffic characteristics, the explanatory variables in heavy vehicle drivers’ lane changing decision were identified. A reliable model has been developed in this thesis to estimate the lane changing behaviour of heavy vehicle drivers. Drivers’ lane changing behaviour has been characterised as a sequence of two stages: the decision to change lanes and the execution of the lane change. Hence, separate models were considered for those two stages of the heavy vehicle drivers’ lane changing behaviour. Fuzzy logic was used to develop a model of the lane changing decision of heavy vehicle drivers. The lane changing decision has been defined as the motivation for selecting either the right adjacent lane (slower lane) or the left adjacent lane (faster lane). Therefore, two separate models were developed for the lane changing decision of heavy vehicle drivers: Lane Changing to Slower Lane (LCSL) and Lane Changing to Faster Lane (LCFL). The explanatory variables in motivating heavy vehicle drivers to move into the slower lane include: the front space gap, the rear space gap, the lag space gap in the right lane and the average speed of the surrounding vehicles in the current lane. The explanatory variables in motivating heavy vehicle drivers to move into the faster lane include: the front relative speed, the lag relative speed in the left lane and the average speeds of the surrounding vehicles in the current lane and the left lane. A triangular membership function was used for all fuzzy sets in the lane changing decision model. The leave-one-out cross-validation method was used to examine the accuracy of the models in estimating the lane changing manoeuvres of heavy vehicle drivers. The obtained results showed that the LCFL model has higher percentage of accurately estimating the heavy vehicle drivers’ lane changing decision. This may be due to the fact that heavy vehicle drivers mainly move into the faster lane to gain speed advantages which could be modelled by the microscopic traffic characteristics of surrounding vehicles in the current and the left lanes. However, the heavy vehicle drivers may have other motivations for moving into the slower lane than only the differences in microscopic traffic characteristics in the current and the right lanes. The speed and acceleration/deceleration profiles of heavy vehicles were analysed in detail from the start to the end of lane changing manoeuvres. The results showed that heavy vehicle drivers maintain an almost constant speed during lane changing execution. They do not accelerate or decelerate to adjust their speed according to the speeds of the surrounding vehicles in the target lane. Subsequently, a simple constant speed model could be assumed for the heavy vehicles during the lane changing execution. Finally, the performance of the heavy vehicle drivers’ lane changing model was examined macroscopically and microscopically using VISSIM (German abbreviation for ‘traffic simulation in cities’) microscopic traffic simulation model. The heavy vehicle lane changing model in VISSIM was substituted with a combination of the fuzzy logic heavy vehicle lane changing decision model and a constant speed lane changing execution model. The traffic measurements obtained from the fuzzy logic model were compared to those obtained from a calibrated VISSIM lane changing model and the actual field observations. The results show that using the fuzzy logic heavy vehicle lane changing model provided more accurate estimates of the macroscopic traffic measurements. The number of heavy vehicle lane changing manoeuvres estimated by the fuzzy logic model was found to be more accurate than the estimates from default lane changing model in VISSIM. The microscopic analysis of the lane changing manoeuvres shows that using the fuzzy logic model more accurately replicated the microscopic lane changing behaviour of the heavy vehicle drivers. In particular, the fuzzy logic model accurately replicates the observed speed profile of heavy vehicles and the observed space gap and speed profiles of the surrounding vehicles during lane changing manoeuvres. The superior performance of the fuzzy logic heavy vehicle drivers’ lane changing model highlights the importance of developing an exclusive lane changing model for heavy vehicle drivers. Employing a purpose built heavy vehicle lane changing model has been shown to increase the accuracy of the microscopic traffic simulation model.