This paper introduces a new hierarchical control method for autonomous electric vehicle (EV) platoons. The method integrates model predictive control (MPC) with fuzzy logic systems to ensure stability in connected vehicle scenarios. The fuzzy MPC framework was developed to address the nonlinearities and interconnections in vehicle dynamics, including electromechanical actuator delay, rolling resistance, tire-to-road interaction, and aerodynamic drag. The multi-level control methodology consists of a high-level fuzzy inference system that adjusts parameters on-the-fly and a low-level MPC that optimizes longitudinal control commands to account for powertrain and brake latencies. The effectiveness of the controller was systematically tested across five distinct scenarios (urban traffic, highway travel, bad weather conditions, cornering maneuver, and mixed platoon) to represent the wide variety of real-world driving situations. Each scenario used an adaptive time-headway policy (THP) that aims to strengthen the system against uncertainties arising from the environment or communication delays, and to outperform traditional fixed-headway policies. Simulation results demonstrate that the fuzzy MPC controller maintains a strict standard of stability, maintains an acceptable distance from safety vehicles, and facilitates coordinated platoon operation across various motion scenarios.
Conventional road traffic safety planning is static and reactive. To address this, an AI‐Driven Dynamic Adaptive Management Framework (AI‐DAMF) for intelligent transportation systems is proposed. The six‐layer architecture integrates multisource data ingestion, knowledge‐guided feature extraction, deep learning‐based pattern recognition, dynamic optimization, IoT‐enabled monitoring, and interactive visualization. Its core innovation is cross‐domain correlation analysis among traffic participants, infrastructure, and operational logistics, which enables proactive risk adjustment and forms a closed‐loop of planning, execution, and evaluation. The primary theoretical contribution is the establishment of a closed‐loop “perception–prediction–optimization–evaluation” paradigm for proactive traffic safety management, shifting the field from reactive incident response to anticipatory risk mitigation. The key practical contribution is a simulation‐validated, scalable framework that transportation engineers and urban planners can directly adopt, achieving a 40% improvement in planning efficiency and a substantial reduction in safety risk identification latency compared to conventional methods. Empirical evaluations further show a decrease in traffic conflict rates under peak‐hour conditions, and the AI‐driven predictor achieves an F1 score of 0.8345 (precision 0.8421 and recall 0.8269). This work delivers a deployable framework that operationalizes the proposed paradigm, ready for regional road traffic safety planning deployment. While the full framework is proposed as an extensible architecture, this study focuses on the implementation and experimental validation of the core risk prediction component.
Travel time is conventionally estimated as a static parametric function of traffic flow, independent of time and space. Such functions are nonunique because two densities correspond to the same flow, and they cannot represent the evolution of congestion. This paper develops travel time as a dynamic state variable at the three classical modelling scales: as an ordinary differential equation for each vehicle in the microscopic setting, as a time-varying scalar field governed by a Hamilton–Jacobi equation coupled to the Lighthill–Whitham–Richards model in the macroscopic setting and as a distribution over position and speed governed by a kinetic transport equation in the mesoscopic setting. Well-posedness of the coupled macroscopic system is established through entropy and viscosity solution theory, the first-in-first-out property is proved for nonnegative speed including complete stops, and numerical schemes are derived for all three scales, with explicit stability conditions for the field formulations. The mesoscopic acceleration is obtained as the first velocity moment of the collision operator, which yields a bounded relaxation law, and the mesoscopic formulation is shown to reduce to the macroscopic one as the speed variance vanishes. The formulation is assessed against classical and existing dynamic travel-time representations, both in a controlled benchmark and against measured vehicle trajectories.
As urban areas in low- and middle-income countries face rising car dependence, congestion, and fragmented transport systems, Mobility-as-a-Service (MaaS) is increasingly proposed as an integrated mobility solution. However, empirical evidence remains limited on MaaS adoption in preimplementation contexts, particularly whether these factors differ between traveler groups. This study examines the determinants of MaaS adoption intention and its behavioral implications among car and public transport users in Greater Accra, Ghana. Drawing on data from a structured survey (n = 1243), structural equation modeling was employed to estimate and compare UTAUT2-based adoption models for private car and public transport users. The results indicate that performance expectancy, effort expectancy, hedonic motivation, and price value significantly influence MaaS adoption intentions. Effort expectancy is the strongest predictor among car users, whereas the price value was the most influential predictor for public transport users; formal multigroup comparisons further indicated that the effects of effort expectancy and price value differed significantly between the groups, highlighting distinct decision-making priorities. Mann–Whitney U test and one-way ANOVA further show that gender, age, and household size significantly shape adoption relationships among public transport users, suggesting a more heterogeneous response pattern, while no significant differences were observed among car users. By examining MaaS adoption in a preimplementation setting, this study advances understanding of behavioral heterogeneity between private car and public transport users while providing context-specific evidence from an underrepresented emerging urban environment. The findings offer practical insights for designing inclusive and scalable MaaS strategies in cities where such systems are yet to be implemented.
