This paper proposes a recurrent neural network based microscopic car following model that is able to accurately capture and predict traffic oscillation. Neural network models have gained increasing popularity in many fields and have been applied in modelling microscopic traffic flow dynamics due to their parameter-free and data-driven nature. We investigate the existing neural network based microscopic car following models, and find out that they are generally accurate in predicting traffic flow dynamics under normal traffic operational conditions. However, they do not maintain accuracy under conditions of traffic oscillation. To bridge this research gap, we first propose four neural network based models and evaluate their applicability to predict traffic oscillation. It is found that, with an appropriate structure and objective function, the recurrent neural network based model has the capability of perfectly re-establishing traffic oscillations and distinguish drivers characteristics. We further compare the proposed model with a classical car following model (Intelligent Driver Model). Based on our case study, the proposed model outperforms the classical car following model in predicting traffic oscillations with different driver characteristics.

A recurrent neural network based microscopic car following model to predict traffic oscillation

The introduction of autonomous (self-driving) and shared autonomous vehicles (AVs and SAVs) will affect travel destinations and distances, mode choice, and congestion. From a traffic perspective, although some congestion reduction may be achieved (thanks to fewer crashes and tighter headways), car-trip frequencies and vehicle miles traveled (VMT) are likely to rise significantly, reducing the benefits of driverless vehicles. Congestion pricing (CP) and road tolls are key tools for moderating demand and incentivizing more socially and environmentally optimal travel choices. This work develops multiple CP and tolling strategies in alternative future scenarios, and investigates their effects on the Austin, Texas network conditions and traveler welfare, using the agent-based simulation model MATSim. Results suggest that, while all pricing strategies reduce congestion, their social welfare impacts differ in meaningful ways. More complex and advanced strategies perform better in terms of traffic conditions and traveler welfare, depending on the development of the mobility landscape of autonomous driving. The possibility to refund users by reinvesting toll revenues as traveler budgets plays a salient role in the overall efficiency of each CP strategy as well as in the public acceptability.

/pdf/congestion-pricing-in-a-world-of-self-driving-vehicles-an-4jhjn27uj2.pdf

Congestion pricing in a world of self-driving vehicles: An analysis of different strategies in alternative future scenarios

We review recent studies on transportation network performance under perturbations. Three representative concepts relating to network performance are covered: reliability, vulnerability, and resilience. With an overview of the definitions and the quantitative indices of these three concepts, we analyse and compare their similarities and differences in the context of transportation. These concepts differ from each other in terms of focus, measurement, and application scenario. Numerical examples are conducted to assess these concepts under different perturbation scenarios. The results indicate their rationale in reflecting network performance under perturbations, yet their outputs differ. Moreover, the relationship among the three concepts is intuitively illustrated by the analysis results.

Performance of transportation network under perturbations: Reliability, vulnerability, and resilience

Despite numerous theoretical studies, practical implementation of congestion pricing is limited mainly due to the low public acceptance. Existing studies in this respect generally focus on a few selected cases where the results need to be further generalized. With the objective of improving public acceptance of congestion pricing, this paper provides a comprehensive overview of the area-based congestion pricing practices. An in-depth analysis of public acceptance is presented using a qualitative case study approach. Results show that for the successful implementation of congestion pricing, a trial and a referendum are valuable but not necessary, and that an interaction-oriented political process may be more desirable. Four influencing factors, i.e. privacy, equity, complexity and uncertainty, are identified to be critical in establishing strong public support. Taking into account these implementation factors, an extended three-step approach is proposed for further improvement of public acceptance toward congestion pricing.

Congestion pricing practices and public acceptance: A review of evidence

This paper proposes a new optimization model for the network design problem of the demand-responsive customized bus (CB). The proposed model consists of two phases: inserting passenger requests dynamically in an interactive manner (dynamic phase) and optimizing the service network statically based on the overall demand (static phase). In the dynamic phase, we propose a hierarchical decision-making model to describe the interactive manner between operator and passengers. The CB network design problem is formulated in a mixed-integer program with the objective of maximizing operator’s revenue. The CB passenger’s travel behavior is measured by a discrete choice model given the trip plan provided by the operator. A dynamic insertion method is developed to address the proposed model in the dynamic phase. For the network design problem in the static phase, the service network is re-optimized based on the confirmed passengers with strict time deviation constraints embedded in the static multi-vehicle pickup and delivery problem. An exact solution method is developed based on the branch-and-bound (B&B) algorithm. Numerical examples are conducted to verify the proposed models and solution algorithms.

A two-phase optimization model for the demand-responsive customized bus network design

This paper investigates the nonlinear distance-based congestion pricing in a network considering stochastic day-to-day dynamics. After an implementation/adjustment of a congestion pricing scheme, the network flows in a certain period of days are not on an equilibrium state, thus it is problematic to take the equilibrium-based indexes as the pricing objective. Therefore, the concept of robust optimization is taken for the congestion toll determination problem, which takes into account the network performance of each day. First, a minimax model which minimizes the maximum regret on each day is proposed. Taking as a constraint of the minimax model, a path-based day to day dynamics model under stochastic user equilibrium (SUE) constraints is discussed in this paper. It is difficult to solve this minimax model by exact algorithms because of the implicity of the flow map function. Hence, a two-phase artificial bee colony algorithm is developed to solve the proposed minimax regret model, of which the first phase solves the minimal expected total travel cost for each day and the second phase handles the minimax robust optimization problem. Finally, a numerical example is conducted to validate the proposed models and methods.

Robust optimization of distance-based tolls in a network considering stochastic day to day dynamics

This study deals with urban transport network flow estimation based on Cellphone Location (CL) and License Plate Recognition (LPR) data. We first propose two methods to filter CL data and extract the spatio-temporal traffic features for a specific road segment. A tailored machine learning approach is developed, including two components: a tangible multi-grained scanning ensemble learning model and a novel two-stage zero-shot learner. The former aims to estimate traffic flow on a single link with both filtered CL data, extracted spatio-temporal traffic features, and LPR data by incorporating the unique merits thereof. The latter is capable of estimating traffic flow on those links with only CL data by considering the spatial features of these links and relevant land-use information. Finally, case studies are analysed to demonstrate the impressive performance of the tailored machine learning approach.

A tailored machine learning approach for urban transport network flow estimation

This paper investigates the day-to-day dynamics model from the perspective of travelers’ actual route choice behaviors, and calibrates and validates the route-based day-to-day dynamics model with the real-world license plate recognition (LPR) data. Due to the highly nonlinear and multi-modal response function in the calibration of the optimization problem, traditional gradient-based nonlinear regression algorithms or other analytical optimization approaches are inapplicable to deal with the calibration work. In this paper, a surrogate-based simulation optimization approach is proposed to deal with the expensive-to-evaluate response function in the day-to-day dynamics calibration work. More specifically, the kriging metamodel is adopted to surrogate the optimization function of the calibration process. With this meta-modeling approach, a sound solution can be achieved with only a few sampling points in a comfortably afforded computation burden, thus giving a valid estimation of the parameters in the day-to-day dynamics model. Finally, a case study based on the real-world LPR data is conducted to validate the proposed model and calibration method.

Surrogate-based simulation optimization approach for day-to-day dynamics model calibration with real data

In this study, a new s-shaped three-parameter (S3) traffic flow model is proposed to represent the relationships between three fundamental variables (i.e., flow, speed, and density) in highway traffic. An s-shaped speed-density function is proposed to capture the speed-density relationship under a wide range of possible densities. A consistent car-following model was derived in terms of the proposed s-shaped speed-density function. Both the S3 macroscopic model and the derived microscopic car-following model were calibrated using real-world traffic data.

Qixiu Cheng

Papers

Congestion pricing practices and public acceptance: A review of evidence

Robust optimization of distance-based tolls in a network considering stochastic day to day dynamics

A tailored machine learning approach for urban transport network flow estimation

Surrogate-based simulation optimization approach for day-to-day dynamics model calibration with real data

An s-shaped three-parameter (S3) traffic stream model with consistent car following relationship