Real-time origin-destination (OD) estimation via anonymous vehicle tracking

This paper presents a road network dynamic origin destination (OD) matrix estimation model using real-time link traffic counts. First, the paper establishes a generalized least squares (GLS) based dynamic OD matrix estimation model, then the dynamic traffic assignment (DTA) technology is used to estimate the link-use proportion matrix. According to the model, a bi-level iterative algorithm is designed to solve the estimation problem. Finally, a case study illustrates that the dynamic OD estimation methodology can trace the variation of network OD matrices well and produce more precise estimation results.

Traffic congestion can be alleviated by infrastructure expansions, however, improving the existing infrastructure using traffic control is more plausible due to the obvious constraints on financial resources and physical space. Independent applications of Adaptive Traffic Signal Control (ATSC) and Ramp Metering (RM) have shown strong potential to effectively alleviate urban traffic congestion by adjusting the signal timings in real-time in response to traffic fluctuations to achieve desirable objectives (e.g., minimize delay at intersections or minimize travel time along freeways). This paper presents the problem formulation and the framework for addressing traffic control problem using an integrated solution combining ATSC and RM. According to that problem definition, an optimal closed-loop approach for ATSC and RM can be designed using a multi-agent reinforcement learning (MARL) approach. Authors' previous studies of MARL on a large-scale simulated network of 59 intersections in the lower downtown core of the City of Toronto for the morning rush hour have shown unprecedented reduction in the average intersection delay up to 39% at the network level, and travel time savings of up 2to 26% along the busiest routes in downtown Toronto. Additionally, authors' studies of MARL applied to a calibrated microsimulation model of the Gardiner Expressway in Toronto, Canada resulted in 50% reduction in total travel time compared with the base case scenario.

Despite its ever-increasing computing power, dynamic origin–destination (OD) estimation in congested networks remains troublesome. In previous research, we have shown that an unbiased estimation requires the calculation of the sensitivity of the link flows to all origin–destination flows, in order to incorporate the effects of congestion spillback. This is, however, computationally infeasible for large-scale networks. To overcome this issue, we propose a hierarchical approach for off-line application that decomposes the dynamic OD estimation procedure in space. The main idea is to perform a more accurate dynamic OD estimation only on subareas where there is congestion spillback. The output of this estimation is then used as input for the OD estimation on the whole network. This hierarchical approach solves many practical and theoretical limitations of traditional OD estimation methods. The main advantage is that different OD estimation method can be used for different parts of the network as necessary. This allows applying more advanced and accurate, but more time-consuming, methods only where necessary. The hierarchical approach is tested on a study network and on a real network. In both cases the proposed methodology performs better than traditional OD estimation approaches, indicating its merit.

A previously developed Deep Reinforcement Learning-based Vehicle Routing (DRL-VR) algorithm aims to be used for providing the shortest Origin-Destination (OD) travel time path in dynamic traffic environment. However, several issues may still arise regarding uncertainty associated with mixed traffic conditions coexisting Automated Vehicles (AV) and Human-driven Vehicles (HV), particularly in a wide-area urban road network. To develop a robust and interoperable route guidance algorithm based on the DRL approach, this study proposes Transfer learning-based deep reinforcement Learning Algorithm for Route Guidance (TLARG). It is an extended framework on the previous approach by incorporating transfer learning scheme that enables the DRL model of TLARG to converge even in a wide-area urban road network. The TLARG is evaluated in terms of OD travel time based on diverse OD trips with different urban road networks, including narrow- and wide-area road networks. This research conducts several evaluation studies based on microscopic traffic simulation experiments. The simulation result shows that the TLARG enables the agent to complete its OD trips not only with flexible routes but also with reductions in travel time depending on given traffic situations irrespective of network type. Furthermore, it demonstrates that the robustness of the proposed approach by measuring the error of Estimated Time of Arrival (ETA) for various OD trips in different urban road networks under the mixed traffic conditions. Such findings suggest that the TLARG has great potential to enhance the punctuality of mobility service by providing robust route guidance, even in the era of coexisting AVs and HVs.

This study proposes an analytical capacity constrained dynamic traffic assignment (DTA) model along with an efficient path-based algorithm. The model can be applied to analyzing dynamic traffic demand management (TDM) strategies, but its specific feature is allowing for an evaluation of advanced traveler information systems (ATIS) within the strategic transportation planning framework. It is an extension of a former DTA model, where each link is given an infinite capacity and is assumed to be completely traversed inside any time interval it is reach by a path. Thereby, the length of time intervals should be very longer than the link travel times, and the link capacity constraints are overlooked. The paper rests on three key ideas to overcome these restrictions: (1) adding path-link fraction variables to the base model, allowing path flows to spread out over time intervals on long links; (2) uniformly dividing each link into smaller parts (segments), so that each part is more likely to be traversed inside a time interval; (3) imposing a dynamic penalty function on each link, thereby the capacity constraint can be included. The proposed DTA algorithm decomposes the augmented model in terms of origin-destination (OD) pairs and departure time intervals, and utilizes a dynamic column generation technique for generating active paths between the OD pairs. The optimal solution to a one-link network demonstrates that the model is able to approximate the dynamic flow propagation over a link with sensible accuracy. Besides, investigation of the results for a small test network reveals that the algorithm performs very well in computing temporal link flows and queuing delays. Finally, numerical experiments on a real life network indicate that the algorithm converges sufficiently fast and provides path information for each time interval. The network is further used to show the algorithm is capable of assessing a dynamic TDM strategy as well as an ATIS system.

In this paper we discuss a dynamic origin–destination (OD) estimation problem that has been used for identifying time-dependent travel demand on a road network. Even though a dynamic OD table is an indispensable data input for executing a dynamic traffic assignment, it is difficult to construct using the conventional OD construction method such as the four-step model. For this reason, a direct estimation method based on field traffic data such as link traffic counts has been used. However, the method does not account for a logical relationship between a travel demand pattern and socioeconomic attributes. In addition, the OD estimation method cannot guarantee the reliability of estimated results since the OD estimation problem has a property named the ‘underdetermined problem.’ In order to overcome such a problem, the method developed in this paper makes use of vehicle trajectory samples with link traffic counts. The new method is applied to numerical examples and shows promising capability for identifying a temporal and spatial travel demand pattern.

In this paper, the estimation of origin-destination (OD) matrices for public traffic passenger flow is given. As the basis of transit scheduling, more accurate OD matrices mean more scientific operations and more rational utilization of staff and vehicles. In order to improve the estimation precision, characteristics of a passenger’s trip, land use around bus stops, and transferring function are considered. Kolmogorov-Smirnov (K-S) Test method is adopted to analyze the trip stops. The result shows that it follows a Poisson distribution. According to passengers boarding and alighting, the OD estimation based on the trip’s character is present. Attracting coefficient and weight are employed to estimate the OD matrices based on the land use around the stops. The stop significant coefficient, which relates to the route number of stations as starting, ending and midway, is also proposed to estimate the OD matrices based on transferring function. Analytic hierarchy process (AHP) is used to combine these three factors, and the last OD matrices are gained. Real data are applied to test the estimation methodology. Good agreement between predicted and observed matrices is found, and the results prove that the method is feasible to heighten accuracy.

Route swapping in dynamic traffic networks

Constant increase in urban road traffic is forcing traffic authorities to confront unavoidable congestion problems these days. This creates new challenges for city planners to generate traffic routes with better designs and improve the existing roads. In order to analyze and improve traffic routes, there is a need to understand movement patterns of major vehicular traffic inside the city. Origin Destination(OD) estimation is one such method for understanding the movement patterns. The objective is to find an optimal OD matrix for city traffic, which is a subtle process for classic algorithms. The solution to this problem is Genetic Algorithm (GA) approach which is a search heuristics to find optimal solution from a set of random solutions. A Genetic Algorithm approach has been developed to find an optimal OD matrix, showing the actual travel patterns of significant number of vehicles inside the city. Once the best OD matrix is obtained, it yields a big opportunity for traffic planners to analyze and improve traffic scenario. Genetic Algorithm can play a major role in solving complex problems. In this paper, one such problem of estimating OD matrix of a city is discussed and implemented. In conjunction to genetic algorithm, Mean Absolute Percentage Error(MAPE) has been used as a fitness criteria for finding the optimal OD matrix.

An adaptive traffic signal controller (ATSC) combined with a connected vehicle (CV) concept uses real-time vehicle trajectory data to regulate green time and has the ability to improve travel time in a signalized corridor. This study introduces an innovative "slow poisoning" cyberattack for a waiting-time-based ATSC algorithm and a corresponding detection strategy. The objectives of this paper are to (1) develop a "slow poisoning" attack generation strategy for an ATSC and (2) develop a prediction-based "slow poisoning" attack detection strategy using a recurrent neural network—i.e., long short-term memory model. We have generated a "slow poisoning" attack modeling strategy using a microscopic traffic simulator—Simulation of Urban Mobility (SUMO)—and used generated data from the simulation to develop both the attack model and detection model. Our analyses revealed that the attack strategy is effective in creating a congestion in an approach and that our detection strategy can capably flag the attack.

Traffic congestion and delays at signalized intersections remain a challenge in urban mobility. To minimize congestion, adaptive traffic signal control systems have been adopted across the United States. However, for these systems to function at their best, systems require an accurate estimation of traffic states, such as traffic density and queue length. This paper implements a relatively novel method, Physics-Informed Neural Networks (PINNs), for real-time traffic density estimation to enhance adaptive intersection signal control. The methodology leverages PINNs, which integrate both data-driven learning and physical constraints imposed by the Lighthill-Whitham-Richards (LWR) macroscopic traffic flow model. A simulation of a four-way intersection was used to generate realistic spatio-temporal traffic data to evaluate the performance of the model. Separate PINNs models, one trained for each incoming approach using simulation-based spatio-temporal density data, provide segment-level density predictions. These real-time density estimates are then employed within an adaptive traffic signal control algorithm designed to adjust green light phasing based on detected demand. The proposed PINNs control strategy was evaluated through simulation, comparing its performance against the Webster traffic timing method using key metrics such as average vehicle delay at the intersection and average vehicle travel time. The findings show an approximate 18% decrease in average vehicle delay and more than a 6% reduction in average travel time when utilizing adaptive traffic control based on the PINNs method for determining density values. These findings highlight the potential of leveraging PINNs for accurate, real-time traffic state estimation, which can effectively inform and improve the performance of adaptive traffic signal control systems, leading to reduced congestion and enhanced mobility at intersections.

This paper proposes a two-stage model for the estimation of origin-destination (OD) demands by the time of day over the year with the use of offline traffic data from the real-time travel information system. In the first stage, a travel time recursive function is proposed to use the offline travel speed data for the investigation of the spatial and temporal relationships between time-dependent OD demands and traffic counts. As such, it is not required to carry out the time-consuming dynamic traffic assignment (DTA) process which is frequently used in the conventional time-dependent OD estimation models. Using the results in the first stage together with the available traffic count data, a least-squares method is adopted to formulate the time-dependent OD demand estimation problem as a quadratic programming model in the second stage. A solution algorithm is adapted for solving the proposed model. Then, the proposed method is easy for implementation in practice. Particularly, when the traffic accident occurs in the network, the estimated time-dependent OD demands can be helpful for understanding the complex travel behavior (e.g., departure time choice) under uncertainty condition. The numerical examples are presented to illustrate the applications of the proposed model.

Distributionally robust origin–destination demand estimation

A heuristic approach to time-dependent origin-destination traffic demand estimation for the development of large-scale dynamic traffic assignment models

In this paper, we study the problem of origin-destination (OD) travel time estimation where the OD input consists of an OD pair and a departure time. We propose a novel neural network based prediction model that fully exploits an important fact neglected by the literature -- for a past OD trip its travel time is usually affiliated with the trajectory it travels along, whereas it does not exist during prediction. At the training phase, our goal is to design novel representations for the OD input and its affiliated trajectory, such that they are close to each other in the latent space. First, we match the OD pairs and their affiliated (historical) trajectories to road networks, and utilize road segment embeddings to represent their spatial properties. Later, we match the timestamps associated with trajectories to time slots and utilize time slot embeddings to represent the temporal properties. Next, we build a temporal graph to capture the weekly and daily periodicity of time slot embeddings. Last, we design an effective encoding to represent the spatial and temporal properties of trajectories. To bind each OD input to its affiliated trajectory, we also encode the OD input into a hidden representation, and make the hidden representation close to the spatio-temporal representation of the trajectory. At the prediction phase, we only use the OD input, get the hidden representation of the OD input, and use it to generate the travel time. Extensive experiments on real datasets show that our method achieves high effectiveness and outperforms existing methods.