Traffic state hysteresis has confounded traffic flow theory for decades. While the literature typically assumes there is a single hysteresis mechanism, we show there are four distinctly different hysteresis mechanisms that have been conflated with each other in the literature and demonstrate another hysteresis mechanism that has not been actively discussed. This paper shows the acceleration hysteresis: (1) should be expected whenever a driver responds with a delay to their leader, (2) should continually perturb the measured traffic state away from the equilibrium fundamental diagram, FD, and (3) is implicit in many traffic flow models. When the state is perturbed, the models continue to employ an unseen equilibrium FD, but because equilibrium states are rarely seen in traffic measurements it should be challenging to establish the equilibrium FD, e.g., potentially having to impute the FD. These findings also suggest a new perspective for higher order models: in addition to capturing driver behavior they could control for how acceleration distorts traffic measurements. After discussing the hysteresis mechanisms, this paper uses naturalistic driving data to explore the two mechanisms that can distort the measured FD: acceleration hysteresis and grade hysteresis. The results suggest that acceleration modulates jam spacing while grade modulates driver response time and the impacts of acceleration. Furthermore, when vehicles come to a stop, the stopped spacing retains the preceding deceleration conditions, resulting in a non-equilibrium state. Thus, one should not use the stopped spacing to calibrate car following models unless they are designed to account for the acceleration hysteresis.
We develop a Cyber-Aware Hybrid Human–Vehicle Model (CH–HVM) for predicting the longitudinal trajectory of a human-driven connected and automated vehicle (CAV) approaching a signalized intersection when driver-facing infrastructure messages may be spoofed. Cyber-induced uncertainty is defined operationally as a mismatch between the physical signal state and the digital cue observed by the driver, such as a corrupted Red Light Countdown (RLCD) or Signal Phase and Timing (SPaT) message; this mismatch enters the stochastic guard through signal and context features and can alter the modeled probability of stopping versus proceeding. The human–machine interaction between driver, vehicle, and infrastructure information is represented as a stochastic hybrid automaton: a data-driven guard maps contextual features into probabilistic stop/pass decisions, and a structured K-phase braking envelope with a bounded residual term governs anomaly-induced motion. We prove that a sufficient safe-gap condition at a valid stopping-regime entry certifies a minimum stop-line margin at the ego vehicle’s first stop, subject to the one-dimensional dynamics, residual bounds, physical admissibility, and execution-lifetime assumptions. For the separate generic complete-dynamics class, we prove conditional existence and uniqueness with no Zeno behavior. We also derive episode-wise model-risk bounds and a finite-sample population bound for a fixed model evaluated on independent inputs. A high-fidelity driving-simulator experiment with spoofed RLCD messages provides descriptive evidence on braking and low-speed responses in 32 trajectory extracts from this single-vehicle intersection setting. Unlike purely physics-based models, CH–HVM embeds a data-driven representation of cyber-affected human decisions and braking behavior; unlike unconstrained machine learning models, it yields closed-form conditional first-stop certificates and statistically justified fixed-model risk bounds for this single-vehicle intersection scenario under adversarial infrastructure messages.
As an emerging mode of trucking operations, platooning refers to virtual coupling of multiple trucks in a convoy, which brings energy and driver labor savings. While centralized truck platooning is expected to yield the most benefit, the issue of truck compliance with a central matching plan has not received much attention despite its practical relevance. Assuming the presence of a platform for centralized truck platooning matching, this study tackles the problem of maximizing total truck utility gain from platooning while minimizing the probability of truck noncompliance with the matching plan. Considering trucks with different origins and destinations, multiple routing possibilities, time window constraints, and different positions in the formed platoon, we first propose a computational procedure to develop the truck platooning matching graph. Then, we characterize the truck investigation process of a matching plan and show the approximate equivalence between the minimum probability of truck noncompliance with a matching plan and the minimum number of blocking pairs. Two integer programming models are subsequently formulated, without and with the possibility of money transfer between trucks in a platoon. To enable compact model formulations, a series of theoretical investigations is conducted to reduce the matching graph by identifying and eliminating edges that will surely not appear in the optimal matching plan. For the model with money transfer which is more challenging to solve, an efficient Benders decomposition-based algorithm is proposed. We further extend the problem to allow a truck to platoon multiple times in its trip, which is plausible for long-haul trips. As the original matching graph is no longer sufficient for multiple-time platooning, a new algorithm is developed with the core idea of constructing connected and time-compatible subgraphs, based on which a modified integer programming model is formulated. Extensive numerical experiments are performed to validate the theoretical investigations, models, and algorithms, providing rich insights about centralized truck platooning matching with compliance considerations. These insights help inform truck platooning deployment toward greater system compliance and efficiency.
Although connected and autonomous vehicles have enabled normal operation in the vast majority of standardized transportation scenarios, they still can be trapped when encountering unexpected situations like accidents or construction zones. In such cases, timely human intervention via remote operators is essential. However, uncertainties in the timing and complexity of intervention tasks, coupled with the high cost of human resources, make it impractical to deploy a large number of operators. This poses a significant challenge in balancing response time and labor cost, particularly in large-scale autonomous fleets with varying levels of intervention urgency. In this study, we propose a pipeline-queue-based framework to enhance the efficiency of human intervention task allocation and enable hierarchical assignment based on priority levels. By incorporating entropic risk constraints and exploiting moment-generating functions, the framework provides tractable risk-sensitive scheduling under assumed task-arrival and service-completion distributions. Numerical experiments using simulated intervention requests show that the proposed approach reduces average waiting times compared with the benchmark model under the tested settings, while illustrating the trade-off between scheduling performance and staffing requirements.
Railway operations require the coordinated use of multi-type resources, such as tracks, platforms, rolling stock, crew, electrical power, and other facilities. Unexpected disruptions frequently occur, leading to conflicts in the planned utilization of multi-type resources (i.e., violations of the available capacity of the resources). In such cases, railway traffic management is essential to resolve conflicts and ensure safe train operations. However, existing methods primarily focus on track conflicts or are highly customized for specific resource types. When disruptions cause conflicts across multi-type resources simultaneously, these methods often fail to provide comprehensive, conflict-free solutions.This paper studies a railway traffic management problem under multi-type resource failures. We present an innovative approach that represents heterogeneous resource utilization in a unified way as time-dependent resource flows, thereby converting different resource failures into a common form of capacity constraints. This enables the integrated consideration of capacity constraints across different resources and the generation of fully conflict-free solutions under a unified framework. Specifically, the time-dependent resource flow (TDRF) formulation is designed to provide a unified representation of multi-type resource conflicts. The Resource Occupancy Table (ROT) and Event-Decision Network (EDN) are proposed to systematically handle multi-type resources, and an integer programming model is developed. Furthermore, an extended Lagrangian relaxation-based heuristic algorithm and a rolling optimization solution framework are designed to efficiently generate conflict-free solutions.Case studies on a 1318-km Chinese high-speed railway line with 23 stations, 139 sidings, 22 power substations, 410 trains, and 161 rolling stock demonstrate that the proposed approach effectively generates conflict-free solutions while accounting for multi-type resources.
