From Models to Mobility Systems: A Survey of Multimodal Dynamic Traffic Assignment

Fokker, Elisabeth; van der Mei, Robert D.; Dugundji, Elenna R. · 2026 · OpenAlex-citations

DOI: 10.1007/s43069-026-00642-1

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This survey addresses the growing need for realistic and scalable Multimodal Dynamic Traffic Assignment (MM-DTA) models as transportation systems evolve toward multimodal, service-oriented, and data-driven frameworks. While traditional DTA models effectively capture temporal variability in congestion and route choices, they often struggle with representing traffic heterogeneity, complex network geometries, and real-world operational features like spillback. The authors identify a significant research gap in existing literature, noting that previous reviews have largely focused on single-modal settings or isolated components like dynamic network loading, rather than the complete assignment process involving multimodal interactions. This paper aims to synthesize recent advances in MM-DTA, highlighting the integration of multiple transportation modes and the emergence of data-driven approaches, particularly deep reinforcement learning (DRL), to bridge the simulation-to-reality gap. The authors conducted a systematic literature search across Scopus and Google Scholar, updated through February 2026, using keywords related to traffic heterogeneity, dynamic assignment, and model characteristics. After applying inclusion criteria that excluded non-road transport and supply chain modeling, 44 publications were retained for detailed analysis. The review categorizes the evolution of MM-DTA into three phases: offline static traffic assignment, offline dynamic traffic assignment, and real-time dynamic traffic assignment. It examines how models incorporate mode-specific attributes such as passenger car equivalents, free-flow speeds, capacity constraints, reaction times, and occupancy rates. The methodology distinguishes between various dynamic network loading implementations, including cell transmission models, link transmission models, and spatial-queue models, while also analyzing equilibrium concepts like user equilibrium and system optimum. The findings reveal that modern MM-DTA models increasingly utilize discrete choice formulations for mode and route selection, coupled with simulation-based or analytical network loading to capture spatio-temporal traffic interactions. The survey highlights specific modeling techniques for heterogeneous traffic, such as using passenger car equivalents to account for space usage differences between cars and trucks, and modeling distinct reaction times to differentiate autonomous vehicles from human-driven vehicles. These distinctions allow models to capture heterogeneous congestion patterns, where different vehicle classes experience varying speeds and densities. The review also identifies that while many models successfully integrate multimodal features, they often rely on static demand assumptions and small-scale networks, limiting their applicability to real-time, large-scale scenarios. The significance of this work lies in its comprehensive synthesis of the transition from traditional, car-centric models to complex, multimodal frameworks. The authors conclude that future research must prioritize scalability for large networks, the incorporation of time-varying and data-driven demand representations, and the integration of real-time traffic information. Specifically, the paper outlines a research agenda centered on leveraging deep reinforcement learning for adaptive, agent-based routing, which offers a promising avenue for capturing real-time traveler decision-making and improving predictive performance in service-oriented mobility systems. This survey serves as a foundational reference for researchers aiming to develop more realistic and actionable traffic assignment models for modern transportation infrastructure.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• mental model of traffic

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Synthesis & Review: research agenda