Analysis and Modeling of Lane Changing Behavior
Lane changing is one of the most basic activities when driving on freeways or arterials. A lane changing maneuver may be classified, depending on the driver’s motivation, as mandatory or discretionary. A mandatory lane change occurs when a driver is trying to move his/her vehicle from its existing lane to the target lane in anticipation of the next turn or to avoid a lane closure downstream. On the contrary, a discretionary lane change occurs when a driver desires a faster speed or wants a greater following distance (i.e., at the driver’s own discretion). Researchers have always assumed that drivers have different decision methodologies and/or risk-taking behavior for these two types of lane changes. This dissertation answers four Research Questions based on analyses of field data: (i) do drivers have different risk-taking behavior when executing a discretionary lane changing maneuver on an arterial street at different times of the day?; (ii) do drivers have different risk-taking behavior between mandatory and discretionary lane changes on freeways?; (iii) do drivers have different risk-taking behavior when executing a mandatory lane changing maneuver at different freeway sites?; and (iv) if the answer to any of the above questions is “yes”, can a lane changing decision model, which has been developed to meet a specific set of driving conditions, be customized to meet another set of driving conditions? To answer each of these questions, the Next Generation SIMulation (NGSIM) vehicle trajectory data sets were used to perform analyses. For the first Research Question, there is enough statistical evidence to conclude that drivers have different risk-taking behavior at different times of the day when making a discretionary lane changing maneuver on an arterial street. The second Research Question asked if there is a statistically significant difference in drivers’ behavior between discretionary and mandatory lane changes. Of the four risk taking parameters tested, there is not enough statistical evidence to suggest that there are significant differences in three of them; however, there is a statistically significant difference in the fourth parameter. As for the third Research Question, again, there is statistical evidence to conclude that drivers have different risk-taking behavior when executing a mandatory lane changing maneuver at different freeway sites. The answers of Research Questions 1 to 3 point to the need for a lane changing model to adapt to different driving environments (i.e. time of the day and locations). The answer to the fourth Research Question has shown that one existing discretionary lane changing model found in the literature does not perform as well when presented with mandatory lane changing data. Therefore, several models have been developed specifically for mandatory lane changes, as it has been proven as part of Research Question 2 that drivers behave differently between the two. An Adaptive Neuro-Fuzzy Inference System (ANFIS), developed as part of this dissertation, is recommended, as it outperforms the existing discretionary lane changing model found in the literature. At a broad scale, the recommended ANFIS model may be incorporated into existing traffic simulation tools (software) to improve the modeling accuracy under different conditions. In the near future, there needs to be a better understanding of how drivers behave when changing lanes, specifically as it relates to automated vehicles. Most vehicles sold today are only partially automated, meaning that the vehicles may assist the driver with speed (e.g., adaptive cruise control), steering (e.g., lane departure warning) and lane changing (e.g., lane change advisory). The results from this dissertation have shown that drivers behave differently based on time-of-day and location. This demonstrates the need for semi- and fully automated vehicles to have a dynamic lane changing model that can adapt to different driving conditions.
Vechione, Matthew Mark, "Analysis and Modeling of Lane Changing Behavior" (2019). ETD Collection for University of Texas, El Paso. AAI13880638.