Optimal Road Traffic Operations for an Increasingly Autonomous and Connected Vehicle Fleet

Osorio, Carolina · 2017 · ROSA P / New England University Transportation Center

archive: archived pipeline: cataloged verified

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

Summary

This report presents a final summary of research funded by the U.S. Department of Transportation, focusing on developing decision-making tools to manage traffic operations in networks with increasing autonomous and connected vehicles. The primary motivation is the need for accurate, real-time calibration of stochastic traffic simulators. As vehicle technology evolves, simulators must accurately reflect current conditions to support applications like real-time congestion routing and incident management. However, calibrating these simulators is computationally expensive because the relationship between input parameters and output measurements is stochastic, nonlinear, and lacks a closed-form expression. Existing online calibration methods often treat simulators as black boxes, relying on numerical gradient approximations that require multiple simulator evaluations, which is inefficient for real-time applications. To address this, the researchers developed a hybrid online calibration algorithm that embeds an analytical approximation, termed a "metamodel," within an Extended Kalman Filter (EKF) framework. This approach replaces the computationally intensive simulator with the metamodel during the online phase. The metamodel consists of two components: a physical term derived from an analytical macroscopic traffic model that relates calibration parameters to measurements using network-specific information, and a functional polynomial term that serves as a local error correction. By using this analytical approximation, the algorithm eliminates the need for numerical gradient approximations, allowing for analytical linearization and significantly reducing computational costs. The simulator is primarily used offline to generate data points for fitting the metamodel parameters via weighted least-squares. The proposed algorithm was validated using the Florian toy road network and the DynaMIT mesoscopic traffic simulator. Experiments involved calibrating origin-destination demand flows using link sensor data under both fixed and time-varying demand scenarios. The method was benchmarked against a standard EKF approach using central finite differences for gradient calculation. Results demonstrated that the proposed algorithm achieved performance comparable to the benchmark in terms of parameter estimation accuracy and robustness to initial estimates. Crucially, it required fewer simulator evaluations, thereby reducing computational cost. The study also evaluated different analytical traffic models for the metamodel, finding that a model based on multinomial logit route choice and the fundamental diagram provided an accurate and scalable approximation. The significance of this work lies in providing a computationally efficient method for online calibration of traffic simulators, which is essential for managing mixed fleets of autonomous and manually driven vehicles. By reducing the computational burden, the algorithm enables more responsive traffic management strategies. The findings support the integration of novel system optimization techniques for autonomous mobility, facilitating large-scale traffic management in connected networks. This approach allows traditional and novel traffic models to accurately represent real-time conditions, enhancing the reliability of predictions and control strategies in increasingly complex transportation systems.

Key finding

The proposed metamodel-based Extended Kalman Filter algorithm achieves calibration accuracy comparable to benchmark methods while requiring fewer simulator evaluations, thereby reducing computational cost.

Methodology

modeling

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. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

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

• situational awareness

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).

• Methodological Resource: validation psychometrics, tool software
• Theoretical Contribution: computational model