Intelligent cruise control (ICC) has the potential to impact both roadway throughput and safety by assisting drivers in maintaining safe headways. This thesis explores this potential through comparisons of ICC to conventional cruise control (CCC) and manual driving. Accordingly, descriptions are given of both CCC and ICC systems. Furthermore, descriptions of ICC evaluation studies and car-following models are presented. The evaluation of ICC is conducted using data collected as part of the Field Operational Test (FOT) performed in Ann Arbor, Michigan. Two levels of analysis are presented in this thesis. The first level of analysis compares the usage of ICC to CCC from a macro level. This study demonstrated that ICC was used more along similar trips. In addition, it was shown that there was no difference in usage of the ON, SET, CANCEL and RESUME buttons. ICC resulted in a higher usage of the ACCEL button and a lower usage of the COAST button compared to CCC. Furthermore, the number of brake interventions while ICC was engaged was higher than CCC. Lastly, the macro-level analysis indicated that there was no difference in the number of near encounters for ICC and CCC. The second analysis makes comparisons at a micro level. The most probable speed, acceleration and headway for each driving mode as well as the probability of using cruise control (based on speed) were determined. The probability of ICC use exceeded CCC use for every freeway speed bin and all but two high-speed arterial speed bins. Finally, a car-following behavior comparison was performed. Manual driving resulted in larger headway values for speeds less than 80 km/h. The ICC speed-headway curve was similar to the CCC speed-headway curve created from high-speed arterial data. The mean headway-speed charts, however, indicated that ICC was more similar to manual driving. Exploration into the specific differences is needed in order to determine the impact of ICC on system safety. / Master of Science

Os simuladores de tráfego são programas computacionais que, através de diversos modelos, tentam simular o tráfego, o comportamento dos motoristas, o desempenho dos veículos, entre outros aspectos que envolvem uma rede viária. Estes modelos precisam ser calibrados para representar as condições de um determinado local. O objetivo da pesquisa foi propor um método de calibração de um microssimulador de tráfego através de dados coletados por estações de monitoramento. O estudo de caso foi realizado através do simulador VISSIM para um trecho do Rodoanel Mário Covas (SP-021), utilizando um algoritmo genético (AG). A calibração envolveu, além dos parâmetros comportamentais dos sub-modelos de car-following e lane-change, o ajuste das distribuições de velocidade desejada dos veículos e um método para simulação do congestionamento. A função fitness do AG foi baseada em três medidas de desempenho: uma que comparava gráficos de fluxo-velocidade simulados e observados e outras duas que comparavam a distribuição do volume de tráfego e o percentual de veículos comerciais por faixa de tráfego. Os resultados mostraram que a medida mais apropriada para a comparação dos gráficos foi a distância de Hausdorff modificada (MHD). A medida MHD também foi fundamental para garantir a ciência do método de simulação de congestionamento de tráfego proposto. O modelo calibrado foi validado usando dados de tráfego coletados em dias diferentes, pela mesma estação de monitoramento. / Traffic simulators are computer programs that, through various models, try to simulate traffic, driver behavior, vehicle performance, and other aspects involved in a road network. These models need calibration to represent local conditions satisfactorily. The objective of the research was to propose a method for the calibration of a traffic microsimulator based on traffic data collected by monitoring stations. To demonstrate the feasibility of the proposed approach, a case study was performed calibrating the simulator VISSIM for a section of Rodoanel Mario Covas (SP-021) using a genetic algorithm (GA). The calibration focused on behavioral parameters for car-following and lane-change submodels, as well as on the desired speed distributions of vehicles and on a method to simulate congestion. The GA fitness function was based on three performance measures: one that compared simulated and observed speed-flow plots, and two that compared the distribution of traffic volume and truck volumes across traffic lanes, respectively. The results showed that the most appropriate measure for comparison of the graphs was the modified Hausdor distance (MHD). MHD was also important to ensure the efficiency of the method used to simulate traffic congestion. The calibrated model was validate using traffic data collected on different days, by the same monitoring station.

Os simuladores de tráfego são programas computacionais que, através de diversos modelos, tentam simular o tráfego, o comportamento dos motoristas, o desempenho dos veículos, entre outros aspectos que envolvem uma rede viária. Estes modelos precisam ser calibrados para representar as condições de um determinado local. O objetivo da pesquisa foi propor um método de calibração de um microssimulador de tráfego através de dados coletados por estações de monitoramento. O estudo de caso foi realizado através do simulador VISSIM para um trecho do Rodoanel Mário Covas (SP-021), utilizando um algoritmo genético (AG). A calibração envolveu, além dos parâmetros comportamentais dos sub-modelos de car-following e lane-change, o ajuste das distribuições de velocidade desejada dos veículos e um método para simulação do congestionamento. A função fitness do AG foi baseada em três medidas de desempenho: uma que comparava gráficos de fluxo-velocidade simulados e observados e outras duas que comparavam a distribuição do volume de tráfego e o percentual de veículos comerciais por faixa de tráfego. Os resultados mostraram que a medida mais apropriada para a comparação dos gráficos foi a distância de Hausdorff modificada (MHD). A medida MHD também foi fundamental para garantir a ciência do método de simulação de congestionamento de tráfego proposto. O modelo calibrado foi validado usando dados de tráfego coletados em dias diferentes, pela mesma estação de monitoramento. / Traffic simulators are computer programs that, through various models, try to simulate traffic, driver behavior, vehicle performance, and other aspects involved in a road network. These models need calibration to represent local conditions satisfactorily. The objective of the research was to propose a method for the calibration of a traffic microsimulator based on traffic data collected by monitoring stations. To demonstrate the feasibility of the proposed approach, a case study was performed calibrating the simulator VISSIM for a section of Rodoanel Mario Covas (SP-021) using a genetic algorithm (GA). The calibration focused on behavioral parameters for car-following and lane-change submodels, as well as on the desired speed distributions of vehicles and on a method to simulate congestion. The GA fitness function was based on three performance measures: one that compared simulated and observed speed-flow plots, and two that compared the distribution of traffic volume and truck volumes across traffic lanes, respectively. The results showed that the most appropriate measure for comparison of the graphs was the modified Hausdor distance (MHD). MHD was also important to ensure the efficiency of the method used to simulate traffic congestion. The calibrated model was validate using traffic data collected on different days, by the same monitoring station.

Nowadays, the location technology in the field of the Intelligent Transformation System (ITS) is used generally. Most of location devices on the cars are low-cost GPS, however, it¡¦s not enough if we want to combine with the safe algorithm. Hence, we present a suit of vehicle collision-avoidance system which combined location technology with Intersection-agent in this thesis. The system uses vehicle sensors and GPS information to calculate in Extend Kalman Filter, in order to get the optimal location information. Furthermore, Map-Matching algorithm is used to match the vehicle location on the right road. As to the driver¡¦s safety, laser range scanner¡¦s data are used in fuzzy algorithm and calculate the safe distance between cars. In the intersection area where accident happened most, we also combine with Intersection-agent system to enhance safety. When moving objects cross through the intersection area, Intersection-agent system would use laser range scanner to find the moving objects¡¦ position and velocity, judging whether they can pass the intersection safely or not. Once it¡¦s not safe, system would send out warning signal to the drivers to brake cars, also, passing the position information to car location system by wireless RS-232 transceiver, to decrease location error and let vehicle¡¦s location precision more accurate. In brief, this thesis combines with vehicle location, wireless transmission, car following warning system and Intersection-agent. And make sure this system we developed can fit in with traffic requirement in many experiments.

The state-of-the-art traffic simulation packages model traffic on a microscopic level. This includes the use of several sets of models that dictate how traffic moves within a transportation network. These models include car-following, gap acceptance, lane-changing and route choice models. The aim of this thesis is to improve the treatment of vehicle dynamics in traffic simulation and, as a result, special attention was paid to car-following models. These models were highlighted because they are largely responsible for capturing a vehicle’s motion and its relevant dynamics in traffic simulation. In order to improve the treatment of vehicle dynamics in traffic simulation, a Dynamic-Interactive-Vehicle (DIV) model was developed. This vehicle model is calibrated with the use of essential vehicle performance specifications that are responsible for the movement of a vehicle in a transportation network. After the calibration process the model is able to accept three inputs from a driver – gas pedal, brake pedal and steering wheel positions. The model then outputs the corresponding longitudinal and latitudinal values which represent the movement of a vehicle along a roadway. The vehicle model will also account for most of the dominant external forces that affect an automobile’s performance along a roadway. This thesis will validate the proposed model by comparing its output from a few performance tests with the performance test results of three passenger cars. The DIV model was validated by comparing the acceleration, braking and steering performance test results of three passenger cars with the output from the DIV model upon performing similar tests. It was found that the DIV model was successful at replicating the two-dimensional vehicle motion.

Though comparisons of simulation models have been conducted, few investigations have examined in detail the logical differences between models. If the output measures of effectiveness are to be interpreted correctly, it is important that the analyst understand some of the underlying logic and assumptions upon which the results are based. An understanding of model logic and its inherent effect on the results will aid the transportation analyst in the application and calibration of a simulation model. In this thesis, the car-following behavior of the CORSIM and INTEGRATION simulation models are examined in significant detail, and its impact on output results explained. In addition, the thesis presents a calibration procedure for the CORSIM sub-model, FRESIM. Currently, FRESIM is calibrated by ad hoc trial-and-error, or by utilizing empirically developed cross-referencing tables. The literature reveals that the relationship between the microscopic input parameters of the CORSIM model, and the macroscopic parameters of capacity is not understood. The thesis addresses this concern. Finally, the thesis compares the INTEGRATION and CORSIM models in freeway and urban environments. The comparison is unique in that the simulated networks were configured such that differences in results could be identified, isolated, and explained. Additionally, the simplified nature of the test networks allowed for the formulation of analytical solutions. The thesis begins by relating steady-state car-following behavior to macroscopic traffic stream models. This is done so that a calibration procedure for the FRESIM (Pipes) car-following model could be developed. The proposed calibration procedure offers an avenue to calibrate microscopic car-following behavior using macroscopic field measurements that can be easily obtained from loop detectors. The calibration procedure, while it does not overcome the inherent shortcomings of the Pipes model, does provide an opportunity to better calibrate the network FRESIM car-following sensitivity factor to existing roadway conditions. The thesis then reports an observed inconsistency in the link-specific car-following sensitivity factor of the FRESIM model. Because calibration of a network on a link-specific basis is key to an accurate network representation, a correction factor was developed that should be applied to the analytically calculated link-specific car-following sensitivity factor. The application of the correction factor resulted in observed saturation flow rates that were within 5% of the desired saturation flow rates. The thesis concludes with a comparison of the CORSIM and INTEGRATION models for transient conditions. As a result of the various intricacies and subtleties that are involved in transient behavior, the comparisons were conducted by running the models on simple networks where analytical solutions to the problem could be formulated. In urban environments, it was observed that the models are consistent in estimates of delay and travel time, and inconsistent in estimates of vehicle stops, stopped delay, fuel consumption, and emissions. Specifically, it was observed that the NETSIM model underestimates the number of vehicle stops in comparison with INTEGRATION and the analytical formulation. It was also observed that the NETSIM vehicles speed and acceleration profiles are characterized by abrupt accelerations and decelerations. These abrupt movements significantly impact stopped time delay and vehicle emissions estimates. Inconsistencies in emissions estimates can also be attributed to differences in the embedded rate tables of each model. In freeway environments for under-saturated conditions, INTEGRATION returned higher values of travel time and delay, and lower values of average speed than the FRESIM model. These results are consistent with the analytical solution, and can be attributed to the speed-flow relationship of each model. In saturated conditions, when the capacity of the bottleneck is equal to the demand volume, the emergent vehicle behavior of the FRESIM model was observed to be inconsistent with the analytical solution. The FRESIM vehicles were observed to dramatically decelerate upon entering a lower-capacity link. This deceleration behavior led to higher travel time and delay time estimates in FRESIM than in INTEGRATION. In over-saturated conditions, longer queue lengths were observed in FRESIM than in INTEGRATION, resulting in slightly higher travel and delay estimates in the FRESIM model. The reason for the discrepancy in queue lengths is unclear, as the network jam density in each model was equivalent. / Master of Science

This research is focused on driver behavior in traffic, especially during car-following situations and safety critical events. Driving behavior is considered as a human decision process in this research which provides opportunities for an artificial driver agent simulator to learn according to naturalistic driving data. This thesis presents two mechine learning methodologies that can be applied to simulate driver naturalistic driving behavior including risk-taking behavior during an incident and lateral evasive behavior which have not yet been captured in existing literature. Two special machine learning approaches Backpropagation (BP) neural network and Neuro-Fuzzy Actor Critic Reinforcement Learning (NFACRL) are proposed to model driver behavior during car-following situation and safety critical events separately. In addition to that, as part of the research, state-of-the-art car-following models are also analyzed and compared to BP neural network approach. Also, driver heterogeneity analyzed by NFACRL method is discussed. Finally, it presents the findings and limitations drawn from each of the specific issues, along with recommendations for further research. / Master of Science

Most phenomena in engineering fields involve physical variables that can potentially be predicted using simple or complex mathematical models. However, traffic engineers and researchers are faced with a complex challenge since they have to deal with the human element. For instance, it can be stated that the biggest challenge facing researchers in the area of car-following theory relates to accounting for the human-in-the-loop while modeling the longitudinal motion of the vehicles. In fact, a major drawback of existing car-following models is that the human-in-the-loop is not modeled explicitly. This is specifically important since the output from car-following models directly impacts several other factors and measures of effectiveness, such as vehicle emissions and fuel consumption levels. The main contribution of this research relates to modeling and incorporating, in an explicit and independent manner, the human-in-the-loop component in car-following theory in such a way that it can be either activated or deactivated depending on if a human driver is in control of the vehicle. That would ensure that a car-following model is able to reflect the different control and autonomy levels that a vehicle could be operated under. Besides that, this thesis offers a better understanding of how humans behave and differ from each other. In fact, through the implementation of explicit parameters representing the human-in-the-loop element, the heterogeneity of human behavior, in terms of driving patterns and styles, is captured. To achieve its contributions, the study starts by modifying the maximum acceleration vehicle-dynamics model by explicitly incorporating parameters that aim to model driver behavior in its expression making it suitable for the representation of typical acceleration behavior. The modified variant of the model is demonstrated to have a flexible shape that allows it to model different types of variations that drivers can generate, and to be superior to other similar models in that it predicts more accurate acceleration levels in all domains. The resulting model is then integrated in the Rakha-Pasumarthy-Adjerid car-following model, which uses a steady-state formulation along with acceleration and collision avoidance constraints to model the longitudinal motion of vehicles. The validation of the model using a naturalistic dataset found that the modified formulation successfully integrated the human behavior component in the model and that the new formulation decreases the modeling error. Thereafter, this dissertation proposes a new car-following model, which we term the Fadhloun-Rakha model. Even though structurally different, the developed model incorporates the key components of the Rakha-Pasumarthy-Adjerid model in that it uses the same steady state formulation, respects vehicle dynamics, and uses very similar collision-avoidance strategies to ensure safe following distances between vehicles. Besides offering a better fit to empirical data, the Fadhloun-Rakha model is inclusive of the following characteristics: (1) it models the driver throttle and brake pedal input; (2) it captures driver variability; (3) it allows for shorter than steady-state following distances when following faster leading vehicles; (4) it offers a much smoother acceleration profile; and (5) it explicitly captures driver perception and control inaccuracies and errors. Through a quantitative and qualitative evaluation using naturalistic data, the new model is demonstrated to outperform other state-of-the-practice car-following models. In fact, the model is proved to result in a significant decrease in the modeling error, and to generate trajectories that are highly consistent with the observed car-following behavior. The final part of this study investigates a case in which the driver is excluded and the vehicles are operating in a connected environment. This section aims to showcase a scenario in which the human-in-the-loop is deactivated through the development of a platooning strategy that governs the motion of connected cooperative multi-vehicle platoons. / Doctor of Philosophy / Even though the study of the longitudinal motion of vehicles spanned over several decades leading to the development of more precise and complex car-following models, an important aspect was constantly overlooked in those models. In fact, due to the complexity of modeling the human-in-the-loop, the vehicle and the driver were almost always assumed to represent a single entity. More specifically, ignoring driver behavior and integrating it to the vehicle allowed avoiding to deal with the challenges related to modeling human behavior. The difficulty of mathematically modeling the vehicle and the driver as two independent components rather than one unique system is due to two main reasons. First, there are numerous car models and types that make it difficult to determine the different parameters impacting the performance of the vehicle as they differ from vehicle to vehicle. Second, different driving patterns exist and the fact that they are mostly dependent on human behavior and psychology makes them very difficult to replicate mathematically. The research presented in this thesis provides a comprehensive investigation of the human-in-the-loop component in car-following theory leading to a better understanding of the human-vehicle interaction. This study was initiated due to the noticeable overlooking of driver behavior in the existing literature which, as a result, fails to capture the effect of human control and perception errors.

This research effort aims to compare car-following models when the models are calibrated to individual drivers with the naturalistic data. The models used are the GHR, Gipps, Intelligent Driver, Velocity Difference, Wiedemann, and the Fritzsche model. This research effort also analyzes the Wiedemann car-following model using car-following periods that occur at different speeds. The Wiedemann car-following model uses thresholds to define the different regimes in car following. Some of these thresholds use a speed parameter, but others rely solely upon the difference in speed between the subject vehicle and the lead vehicle. This research effort also reconstructs the Wiedemann car-following model for truck driver behavior using the Naturalistic Truck Driving Study's (NTDS) conducted by Virginia Tech Transportation Institute. This Naturalistic data was collected by equipping 9 trucks with various sensors and a data acquisition system. This research effort also combines the Wiedemann car-following model with the GHR car-following model for trucks using The Naturalistic Truck Driving Study's (NTDS) data. / Master of Science

Microscopic traffic simulation software have gained significant popularity and are widely used both in industry and research mainly because of the ability of these tools to reflect the dynamic nature of the transportation system in a stochastic fashion. To better utilize these software, it is necessary to understand the underlying logic and differences between them. A Car-following model is the core of every microscopic traffic simulation software. In the context of this research, the thesis develops procedures for calibrating the steady-state car-following models in a number of well known microscopic traffic simulation software including: CORSIM, AIMSUN, VISSIM, PARAMICS and INTEGRATION and then compares the VISSIM and INTEGRATION software for the modeling of traffic signalized approaches. The thesis presents two papers. The first paper develops procedures for calibrating the steady-state component of various car-following models using macroscopic loop detector data. The calibration procedures are developed for a number of commercially available microscopic traffic simulation software, including: CORSIM, AIMSUN2, VISSIM, Paramics, and INTEGRATION. The procedures are then applied to a sample dataset for illustration purposes. The paper then compares the various steady-state car-following formulations and concludes that the Gipps and Van Aerde steady-state car-following models provide the highest level of flexibility in capturing different driver and roadway characteristics. However, the Van Aerde model, unlike the Gipps model, is a single-regime model and thus is easier to calibrate given that it does not require the segmentation of data into two regimes. The paper finally proposes that the car-following parameters within traffic simulation software be link-specific as opposed to the current practice of coding network-wide parameters. The use of link-specific parameters will offer the opportunity to capture unique roadway characteristics and reflect roadway capacity differences across different roadways. Second, the study compares the logic used in both the VISSIM and INTEGRATION software, applies the software to some simple networks to highlight some of the differences/similarities in modeling traffic, and compares the various measures of effectiveness derived from the models. The study demonstrates that both the VISSIM and INTEGRATION software incorporate a psycho-physical car-following model which accounts for vehicle acceleration constraints. The INTEGRATION software, however uses a physical vehicle dynamics model while the VISSIM software requires the user to input a vehicle-specific speed-acceleration kinematics model. The use of a vehicle dynamics model has the advantage of allowing the model to account for the impact of roadway grades, pavement surface type, pavement surface condition, and type of vehicle tires on vehicle acceleration behavior. Both models capture a driver's willingness to run a yellow light if conditions warrant it. The VISSIM software incorporates a statistical stop/go probability model while current development of the INTEGRATION software includes a behavioral model as opposed to a statistical model for modeling driver stop/go decisions. Both software capture the loss in capacity associated with queue discharge using acceleration constraints. The losses produced by the INTEGRATION model are more consistent with field data (7% reduction in capacity). Both software demonstrate that the capacity loss is recovered as vehicles move downstream of the capacity bottleneck. With regards to fuel consumption and emission estimation the INTEGRATION software, unlike the VISSIM software, incorporates a microscopic model that captures transient vehicle effects on fuel consumption and emission rates. / Master of Science