Hengxing Cai, Jinhan Dong, Jingjun Tan, Jingcheng Deng, Sihang Li, Zhifeng Gao, Haidong Wang, Zicheng Su, Agachai Sumalee, Renxin Zhong. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Balancing the supply and demand for ride-sourcing companies is a challenging issue, especially with real-time requests and stochastic traffic conditions of large-scale congested road networks. To tackle this challenge, this article proposes a robust and scalable approach that integrates reinforcement learning (RL) and a centralized programming (CP) structure to promote real-time taxi operations. Both real-time order matching decisions and vehicle relocation decisions at the microscopic network scale are integrated within a Markov decision process framework. The RL component learns the decomposed state-value function, which represents the taxi drivers' experience, the off-line historical demand pattern, and the traffic network congestion. The CP component plans nonmyopic decisions for drivers collectively under the prescribed system constraints to explicitly realize cooperation. Furthermore, to circumvent sparse reward and sample imbalance problems over the microscopic road network, this article proposed a temporal-difference learning algorithm with prioritized gradient descent and adaptive exploration techniques. A simulator is built and trained with the Manhattan road network and New York City yellow taxi data to simulate the real-time vehicle dispatching environment. Both centralized and decentralized taxi dispatching policies are examined with the simulator. This case study shows that the proposed approach can further improve taxi drivers' profits while reducing customers' waiting times compared to several existing vehicle dispatching algorithms.

Serious roadside pollution in congested urban areas is an ongoing problem in many densely populated cities. While the current control measures on traffic pollution have reduced particulate matter, new solutions by emerging vehicular technologies can help protect citizens from exhaust-gas emissions. The electric vehicle (EV) is a promising solution to alleviate traffic-induced pollution in urban areas. However, traffic flow will be mixed with fuel vehicles (FVs) and EVs before the FV would be phased out. In this paper, traffic management by road pricing is introduced to reduce the tailpipe emission of the protected area by imposing environmental capacity constraints for network traffic mixed with EVs and FVs. Second-best toll pricing schemes are formulated as side constrained user equilibrium problems for both fixed demand and elastic demand cases. Both the tailpipe emission of FVs and the energy consumption of EVs are assumed to depend nonlinearly on network traffic conditions. Although the EVs do not contribute to the tailpipe emission, all vehicles contribute to congestion externality that induces more emission of FVs. Therefore, both types of vehicles are charged, but the toll on FVs is significantly higher than that on EVs. A new projected dynamics based algorithm is introduced to solve the toll from the Lagrange multiplier associated with the environmental constraint apart from the equilibrium flow. Numerical examples are conducted to evaluate the equilibrium cost and toll, and to analyze the impacts of EV penetration rate on the pricing scheme.

This paper proposes an alternative method for bus passenger waiting time estimation using passive Wi-Fi data. With the underlying mechanism for Wi-Fi communication, the presence of Wi-Fi enabled devices in a particular area can be passively discoverable. The mobile devices carried by bus passengers can be exploited to estimate passenger waiting time at a bus stop without the passengers' direct participation. Passenger waiting time estimation using such opportunistic data is challenging due to the particular characteristics of the Wi-Fi data collected from bus stop environments. This paper proposes a methodology to handle massive noise in Wi-Fi data and identify the potential Wi-Fi records which are derived from passengers' devices. The filtered data can then be used to estimate passenger waiting time. The Wi-Fi data collected from a bus stop in Hong Kong are used as a case study for evaluating the proposed system. The practicality is investigated in terms of estimation accuracy and insightful analysis.

This paper investigates the dynamic system optimum (DSO) problem with simultaneous route and departure time assignments for a general traffic network partitioned into multiple regions. Regional traffic congestion is modeled with a well-defined macroscopic fundamental diagram (MFD) mapping the trip completion rate to the vehicular accumulation. To overcome the limitation of inconsistent flow propagation between region boundaries and the corresponding travel time, the state-dependent regional travel time function is explicitly incorporated in the flow propagation of the conventional MFD dynamics. From a systems perspective, the traffic dynamics within a region can be regarded as a dynamic system with an endogenous time-varying delay depending on the system state. Equilibrium condition for the DSO problem is analytically derived through the lens of Pontryagin minimum principle and is compared against the static SO counterpart. The structure of path specific marginal cost is analyzed regarding the path travel cost and early-late penalty function. In contrast to existing analytical methods, the proposed method is applicable for general MFD systems without linearization of the MFD dynamics. Neither approximation of the equilibrium solution nor constant regional delay assumption is required. Numerical examples are conducted to illustrate the characteristics of DSO traffic equilibrium and the corresponding marginal cost together with other dynamic external costs.

This paper proposes a vehicle reidentification (VRI) system with self-adaptive time windows to estimate the mean travel time for each time period on the freeway under traffic demand and supply uncertainty. To capture the traffic dynamics in real-time application, interperiod adjusting based on the exponential smoothing technique is introduced to define an appropriate time-window constraint for the VRI system. In addition, an intraperiod adjusting technique is also employed to handle the nonpredictable traffic congestion. To further reduce the negative effect caused by the mismatches, a postprocessing technique, including thresholding and stratified sampling, is performed on the travel time data derived from the VRI system. Several representative tests are carried out to evaluate the performance of the proposed VRI against potential changes in traffic conditions, e.g., recurrent traffic congestion, freeway bottlenecks, and traffic incidents. The results show that this method can perform well under traffic demand and supply uncertainty.

As travel times in road networks are dynamic and uncertain, it is difficult and time-consuming to search for the least expected time path in large-scale networks. This paper addresses the problem of finding the least expected time path in stochastic time-dependent (STD) road networks. A stochastic travel speed model is proposed to represent STD link travel times. It is proved that the link travel times in STD networks satisfy the stochastic first-in-first-out (S-FIFO) property. Based on this S-FIFO property, an efficient multicriteria A* algorithm is proposed to exactly determine the least expected time path in STD networks. Computational results using several large-scale road networks show that the proposed algorithm has a significant computational advantage over existing solution algorithms without the S-FIFO property.

The stochastic cell transmission model (SCTM) was originally developed for stochastic dynamic traffic state modeling under several assumptions, e.g., the independent/uncorrelated assumption of the underlying stochastic processes governing demand and supply uncertainties. However, traffic flow, by nature, is correlated in both spatial and temporal domains due to its dynamics, similar environmental conditions and human behaviors. The independent assumption in the original SCTM framework may prevent the model from a broad range of applications, e.g., short-term traffic state prediction. In this paper, the SCTM framework is extended to consider the spatial-temporal correlation of traffic flow and to support short-term traffic state prediction. First, a multivariate normal distribution (MND)-based best linear predictor is adopted as an auxiliary dynamical system to the original SCTM to forecast boundary variables and/or supply functions. The predicted boundary variables and supply functions are taken as inputs to the SCTM to perform short-term traffic state prediction. The independent assumption of the SCTM is relaxed by incorporating the covariance structure calibrated from the spatial correlation analysis for probabilistic traffic state evaluation. For real-time application purposes, prediction is conducted in a rolling horizon manner, which is useful for adjusting the predicted traffic state using real-time measurements. The proposed traffic state prediction framework is validated by empirical studies that demonstrate the effectiveness of the proposed method.

This article proposes an efficient solution algorithm to aid travelers' route choice decisions in road network with travel time uncertainty, in the context of advanced traveler information systems (ATIS). In this article, the travel time of a link is assumed to be spatially correlated only to the neighboring links within a local ‘impact area.’ Based on this assumption, the spatially dependent reliable shortest path problem (SD-RSPP) is formulated as a multicriteria shortest path-finding problem. The dominant conditions for the SD-RSPP are established in this article. A new multicriteria A* algorithm is proposed to solve the SD-RSPP in an equivalent two-level hierarchical network. A case study using real-world data shows that link travel times are, indeed, only strongly correlated within the local impact areas; and the proposed limited spatial dependence assumption can well approximate path travel time variance when the size of the impact area is sufficiently large. Computational results demonstrate that the size of the impact area would have a significant impact on both accuracy and computational performance of the proposed solution algorithm.