Perspectives

Overview

These reviews and perspectives revisit established approaches, synthesize emerging evidence, and identify future research priorities across transportation data, modeling, operations, and system risk:

• Large language models are emerging as a potential bridge between semantic understanding, reasoning, and decision-making in autonomous driving. We reviewed their roles in both modular and end-to-end driving systems, assessed how they may address long-tail scenarios and improve interpretability, and outlined the technical, safety, security, and societal challenges that must be resolved on the path toward higher-level autonomous driving^[1].
• Urban transportation increasingly depends on tightly coupled modes, yet resilience research has often examined each network in isolation. We organized the field around network modeling, resilience evaluation, and resilience optimization, compared mono-modal and multi-modal approaches, and incorporated emerging modes such as shared mobility and urban air mobility^[2].
• Space weather can disrupt navigation, communications, power systems, and transportation operations, yet its cross-modal effects had not been comprehensively synthesized. We reviewed impacts on air, maritime, rail, and ground transportation, connected the underlying disruption mechanisms to historical evidence, and outlined forecasting, mitigation, and resilient-infrastructure priorities^[3].
• Transportation vulnerability and resilience research spans overlapping concepts and disconnected analytical traditions. We clarified these concepts from supply-and-demand perspectives, organized studies into topological and system-structure approaches, and summarized the growing roles of simulation, mathematical modeling, and data-driven analysis^[4].
• Research on automated-vehicle traffic flow had grown rapidly around heterogeneous assumptions and largely unvalidated models. We systematically reviewed microscopic, mesoscopic, and macroscopic models, linked automated-driving strategies with traffic dynamics and management, and highlighted the need for field data and more realistic mixed-traffic assumptions^[5].
• The availability of vehicle trajectories, especially NGSIM data, reshaped empirical traffic-flow research but also produced fragmented findings across scales. We revisited 15 years of trajectory-based traffic-flow studies, unified new phenomena and models from microscopic to macroscopic levels, and identified priorities for higher-quality data and theory development^[6].
• Vehicle-mounted and mobile sensors create new opportunities for observing travelers, vehicles, and surrounding traffic at scale. We synthesized vehicle sensor data research across data quality, human behavior, transportation operations, emissions, and policy, showing how these data support modeling, analysis, control, and management^[7].
• Many route-guidance strategies had been evaluated only in idealized symmetric networks, leaving their performance under asymmetric conditions uncertain. By comparing eight strategies in an asymmetric two-route network, we showed that only mean-velocity feedback consistently approximated user optimality and clarified why other feedback measures failed^[8].

Publications