No description available.

No description available.

Bluetooth Data collection technique is recently proven as a reliable data collection technique that provides the opportunity to modify traditional methodologies to improve system performance. Actual volume in the network is a result of the timing plans which are designed and modified based on the volume which is generated using existing timing plans in the system. This interdependency between timing plan and volume on the network is a dynamic process and should be captured to obtain actual traffic states in the network. The current practice is to calculate synthetic origin destination information based on detector volume that doesn't necessarily represent the volume scenario accurately. The data from Bluetooth technology can be utilized to calculate dynamic volume on the network which can be further used as input for signal timing design. Application of dynamic volume improves the system performance by providing the actual volume in system to design optimal timing plans. This thesis proposes a framework that can be used to integrate data obtained from the Bluetooth technology with the traditional methods to design timing plans. The proposed methodology utilizes the origin destination information obtained from Bluetooth data, detector data, characteristics of intersections such as number of lanes, saturation flow rate and existing timing plans as a basis for the calculation of the dynamic volume for the various movements at each intersection. The study shows that using the Bluetooth based OD matrix to calculate accurate dynamic volumes results in better system performance compared to the traditional way of using the static detector volume alone. / Master of Science

This dissertation develops and evaluates a new adaptive traffic signal control system for arterials. This control system is based on reinforcement learning, which is an important research area in distributed artificial intelligence and has been extensively used in many applications including real-time control. In this dissertation, a systematic comparison between the reinforcement learning control methods and existing adaptive traffic control methods is first presented from the theoretical perspective. This comparison shows both the connections between them and the benefits of using reinforcement learning. A Neural-Fuzzy Actor-Critic Reinforcement Learning (NFACRL) method is then introduced for traffic signal control. NFACRL integrates fuzzy logic and neural networks into reinforcement learning and can better handle the curse of dimensionality and generalization problems associated with ordinary reinforcement learning methods. This NFACRL method is first applied to isolated intersection control. Two different implementation schemes are considered. The first scheme uses a fixed phase sequence and variable cycle length, while the second one optimizes phase sequence in real time and is not constrained to the concept of cycle. Both schemes are further extended for arterial control, with each intersection being controlled by one NFACRL controller. Different strategies used for coordinating reinforcement learning controllers are reviewed, and a simple but robust method is adopted for coordinating traffic signals along the arterial. The proposed NFACRL control system is tested at both isolated intersection and arterial levels based on VISSIM simulation. The testing is conducted under different traffic volume scenarios using real-world traffic data collected during morning, noon, and afternoon peak periods. The performance of the NFACRL control system is compared with that of the optimized pre-timed and actuated control. Testing results based on VISSIM simulation show that the proposed NFACRL control has very promising performance. It outperforms optimized pre-timed and actuated control in most cases for both isolated intersection and arterial control. At the end of this dissertation, issues on how to further improve the NFACRL method and implement it in real world are discussed.

With the rapid growth in modern cities and congestion on major freeways and local streets, public transit services have become more and more important for urban transportation. As an important component of Intelligent Transportation Systems (ITS), Transit Signal Priority (TSP) systems have been extensively studied and widely implemented to improve the quality of transit service by reducing transit delay. The focus of this research is on the development of a platform with the physical queue representation that can be employed to evaluate and/or improve TSP strategies with the consideration of the interaction between transit vehicles and queues at the intersection.This dissertation starts with deterministic analyses of TSP systems based on a physical queue model. A request oriented TSP decision process is then developed which incorporates a set of TSP decision regions defined on a time-space diagram with the physical queue representation. These regions help identify the optimal detector location, select the appropriate priority control strategy, and handle the situations with multiple priority requests. In order to handle uncertainties in TSP systems arising in bus travel time and dwell time estimation, a type-2 fuzzy logic forecasting system is presented and tested with field data. Type-2 fuzzy logic is very powerful in dealing with uncertainty. The use of Type-2 fuzzy logic helps improve the performance of TSP systems. The last component of the dissertation is the development of a Colored Petri Net (CPN) model for TSP systems. With CPN tools, computer simulation can be performed to evaluate various TSP control strategies and the decision process. Examples for demonstrating the process of implementing the green extension strategy and the proposed TSP decision process are presented in the dissertation. The CPN model can also serve as an interface between the platform developed in this dissertation and the implementation of the control strategies at the controller level.

Um melhor uso da infraestrutura da rede de transporte é um ponto fundamental para atenuar os efeitos dos congestionamentos no trânsito. Este trabalho utiliza aprendizado por reforço multiagente (MARL) para melhorar o uso da infraestrutura e, consequentemente, mitigar tais congestionamentos. A partir disso, diversos desafios surgem. Primeiro, a maioria da literatura assume que os motoristas aprendem (semáforos não possuem nenhum tipo de aprendizado) ou os semáforos aprendem (motoristas não alteram seus comportamentos). Em segundo lugar, independentemente do tipo de classe de agentes e do tipo de aprendizado, as ações são altamente acopladas, tornando a tarefa de aprendizado mais difícil. Terceiro, quando duas classes de agentes co-aprendem, as tarefas de aprendizado de cada agente são de natureza diferente (do ponto de vista do aprendizado por reforço multiagente). Finalmente, é utilizada uma modelagem microscópica, que modela os agentes com um alto nível de detalhes, o que não é trivial, pois cada agente tem seu próprio ritmo de aprendizado. Portanto, este trabalho não propõe somente a abordagem de co-aprendizado em agentes que atuam em ambiente compartilhado, mas também argumenta que essa tarefa precisa ser formulada de forma assíncrona. Além disso, os agentes motoristas podem atualizar os valores das ações disponíveis ao receber informações de outros motoristas. Os resultados mostram que a abordagem proposta, baseada no coaprendizado, supera outras políticas em termos de tempo médio de viagem. Além disso, quando o co-aprendizado é utilizado, as filas de veículos parados nos semáforos são menores. / A better use of transport network infrastructure is a key point in mitigating the effects of traffic congestion. This work uses multiagent reinforcement learning (MARL) to improve the use of infrastructure and, consequently, to reduce such congestion. From this, several challenges arise. First, most literature assumes that drivers learn (traffic lights do not have any type of learning) or the traffic lights learn (drivers do not change their behaviors). Second, regardless of the type of agent class and the type of learning, the actions are highly coupled, making the learning task more difficult. Third, when two classes of agents co-learn, the learning tasks of each agent are of a different nature (from the point of view of multiagent reinforcement learning). Finally, a microscopic modeling is used, which models the agents with a high level of detail, which is not trivial, since each agent has its own learning pace. Therefore, this work does not only propose the co-learnig approach in agents that act in a shared environment, but also argues that this taks needs to be formulated asynchronously. In addtion, driver agents can update the value of the available actions by receiving information from other drivers. The results show that the proposed approach, based on co-learning, outperforms other policies regarding average travel time. Also, when co-learning is use, queues of stopped vehicles at traffic lights are lower.

Adaptive traffic signal control is the process by which the timing of a traffic signal is continuously adjusted based on the changing arrival patterns of vehicles at an intersection, usually with the goal of optimizing a given measure of effectiveness. Herein, a methodology is developed in which the characteristics of a traffic signal cycle are optimized at the conclusion of every phase based on the arrival times of vehicles to an intersection, using stopped delay as the measure of effectiveness. This optimization is solved using metaheuristic search procedures, namely tabu search, and embedded in an algorithm in which current vehicle arrival times are detected, arrival patterns over a specified horizon are predicted, the traffic signal timing is optimized, and the timings are sent to a traffic signal controller. The methodology is shown to provide improvement in performance for a number of intersection configurations and traffic regimes over traditional forms of traffic signal control, and the metaheuristic search is demonstrated to significantly reduce the computation time for a solution as compared with other search procedures. / text

Signal control systems have experienced tremendous development both in hardware and in control strategies in the past 50 years since the advent of the first electronic traffic signal control device. The state-of-art real-time signal control strategies rely heavily on infrastructure-based sensors, including in-pavement or video based loop detectors for data collection. With the emergence of connected vehicle technology, mobility applications utilizing vehicle to infrastructure (V2I) communications enable the intersection to acquire a much more complete picture of the nearby vehicle states. Based on this new source of data, traffic controllers should be able to make "smarter" decisions. This dissertation investigates the traffic signal control strategies in a connected vehicle environment considering mobility as well as safety. A system architecture for connected vehicle based signal control applications under both a simulation environment and in the real world has been developed. The proposed architecture can be applied to applications such as adaptive signal control, signal priority including transit signal priority (TSP), freight signal priority (FSP), emergency vehicle preemption, and integration of adaptive signal control and signal priority. Within the framework, the trajectory awareness of connected vehicles component processes and stores the connected vehicle data from Basic Safety Message (BSM). A lane level intersection map that represents the geometric structure was developed. Combined with the map and vehicle information from BSMs, the connected vehicles can be located on the map. Some important questions specific to connected vehicle are addressed in this component. A geo-fencing area makes sure the roadside equipment (RSE) receives the BSM from only vehicles on the roadway and within the Dedicated Short-range Communications (DSRC) range. A mechanism to maintain anonymity of vehicle trajectories to ensure privacy is also developed. Vehicle data from the trajectory awareness of connected vehicles component can be used as the input to a real-time phase allocation algorithm that considers the mobility aspect of the intersection operations. The phase allocation algorithm applies a two-level optimization scheme based on the dual ring controller in which phase sequence and duration are optimized simultaneously. Two objective functions are considered: minimization of total vehicle delay and minimization of queue length. Due to the low penetration rate of the connected vehicles, an algorithm that estimates the states of unequipped vehicles based on connected vehicle data is developed to construct a complete arrival table for the phase allocation algorithm. A real-world intersection is modeled in VISSIM to validate the algorithms. Dangerous driving behaviors may occur if a vehicle is trapped in the dilemma zone which represents one safety aspect of signalized intersection operation. An analytical model for estimating the number of vehicles in dilemma zone (NVDZ) is developed on the basis of signal timing, arterial geometry, traffic demand, and driving characteristics. The analytical model of NVDZ calculation is integrated into the signal optimization to perform dilemma zone protection. Delay and NVDZ are formulated as a multi-objective optimization problem addressing efficiency and safety together. Examples show that delay and NVDZ are competing objectives and cannot be optimized simultaneously. An economic model is applied to find the minimum combined cost of the two objectives using a monetized objective function. In the connected vehicle environment, the NVDZ can be calculated from connected vehicle data and dilemma zone protection is integrated into the phase allocation algorithm. Due to the complex nature of traffic control systems, it is desirable to utilize traffic simulation in order to test and evaluate the effectiveness and safety of new models before implementing them in the field. Therefore, developing such a simulation platform is very important. This dissertation proposes a simulation environment that can be applied to different connected vehicle related signal control applications in VISSIM. Both hardware-in-the-loop (HIL) and software-in-the-loop (SIL) simulation are used. The simulation environment tries to mimic the real world complexity and follows the Society of Automotive Engineers (SAE) J2735 standard DSRC messaging so that models and algorithms tested in the simulation can be directly applied in the field with minimal modification. Comprehensive testing and evaluation of the proposed models are conducted in two simulation networks and one field intersection. Traffic signal priority is an operational strategy to apply special signal timings to reduce the delay of certain types of vehicles. The common way of serving signal priority is based on the "first come first serve" rule which may not be optimal in terms of total priority delay. A priority system that can serve multiple requests with different priority levels should perform better than the current method. Traditionally, coordination is treated in a different framework from signal priority. However, the objectives of coordination and signal priority are similar. In this dissertation, adaptive signal control, signal priority and coordination are integrated into a unified framework. The signal priority algorithm generates a feasible set of optimal signal schedules that minimize the priority delay. The phase allocation algorithm considers the set as additional constraints and tries to minimize the total regular vehicle delay within the set. Different test scenarios including coordination request, priority vehicle request and combination of coordination and priority requests are developed and tested.

Dissertation for the degree of Doctor of Science in Technology--Helsinki University of Technology, Espoo, 2002. / "ISSN 0781-5816." Includes bibliographical references (p. 65-71). Available online as a PDF file via the World Wide Web.

Um melhor uso da infraestrutura da rede de transporte é um ponto fundamental para atenuar os efeitos dos congestionamentos no trânsito. Este trabalho utiliza aprendizado por reforço multiagente (MARL) para melhorar o uso da infraestrutura e, consequentemente, mitigar tais congestionamentos. A partir disso, diversos desafios surgem. Primeiro, a maioria da literatura assume que os motoristas aprendem (semáforos não possuem nenhum tipo de aprendizado) ou os semáforos aprendem (motoristas não alteram seus comportamentos). Em segundo lugar, independentemente do tipo de classe de agentes e do tipo de aprendizado, as ações são altamente acopladas, tornando a tarefa de aprendizado mais difícil. Terceiro, quando duas classes de agentes co-aprendem, as tarefas de aprendizado de cada agente são de natureza diferente (do ponto de vista do aprendizado por reforço multiagente). Finalmente, é utilizada uma modelagem microscópica, que modela os agentes com um alto nível de detalhes, o que não é trivial, pois cada agente tem seu próprio ritmo de aprendizado. Portanto, este trabalho não propõe somente a abordagem de co-aprendizado em agentes que atuam em ambiente compartilhado, mas também argumenta que essa tarefa precisa ser formulada de forma assíncrona. Além disso, os agentes motoristas podem atualizar os valores das ações disponíveis ao receber informações de outros motoristas. Os resultados mostram que a abordagem proposta, baseada no coaprendizado, supera outras políticas em termos de tempo médio de viagem. Além disso, quando o co-aprendizado é utilizado, as filas de veículos parados nos semáforos são menores. / A better use of transport network infrastructure is a key point in mitigating the effects of traffic congestion. This work uses multiagent reinforcement learning (MARL) to improve the use of infrastructure and, consequently, to reduce such congestion. From this, several challenges arise. First, most literature assumes that drivers learn (traffic lights do not have any type of learning) or the traffic lights learn (drivers do not change their behaviors). Second, regardless of the type of agent class and the type of learning, the actions are highly coupled, making the learning task more difficult. Third, when two classes of agents co-learn, the learning tasks of each agent are of a different nature (from the point of view of multiagent reinforcement learning). Finally, a microscopic modeling is used, which models the agents with a high level of detail, which is not trivial, since each agent has its own learning pace. Therefore, this work does not only propose the co-learnig approach in agents that act in a shared environment, but also argues that this taks needs to be formulated asynchronously. In addtion, driver agents can update the value of the available actions by receiving information from other drivers. The results show that the proposed approach, based on co-learning, outperforms other policies regarding average travel time. Also, when co-learning is use, queues of stopped vehicles at traffic lights are lower.