Experimenting With an Efficient Driver Behavior Dynamical Model Applicable to Simulated Lane Changing Tasks

Jirgl, Miroslav; Mihálik, Ondrej; Boujenfa, Sabrina; Bradáč, Zdeněk; Fiedler, Petr · 2024 · OpenAlex-citations

DOI: 10.1109/access.2024.3451622

archive: archived pipeline: cataloged verified

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

Summary

This study addresses the challenge of modeling human driver behavior during simulated lane-changing tasks, aiming to identify a dynamical model that balances precision with simplicity. Motivated by the need for interpretable models in human-machine shared control systems, the authors focus on linear cybernetic approaches based on McRuer’s crossover theory, rather than complex non-linear or neural network models. The primary objective is to determine the optimal structure of a single-loop transfer function using data-driven methods, specifically cross-validation, to ensure the model contains the minimal number of parameters necessary for accurate approximation. The experimental design utilized an in-house developed car driving simulator built on Unreal Engine 4, featuring a Logitech G920 steering wheel and a high-resolution display. Data were collected from 92 active drivers performing a "step response" scenario, where participants executed lane changes at a constant speed of 90 km/h in response to visual cues. This setup allowed for high repeatability and controlled conditions, isolating lateral control actions from longitudinal dynamics. The researchers addressed non-uniform sampling rates inherent to the simulation by resampling data using Cubic Smoothing Splines. Model selection was conducted using a leave-one-out cross-validation technique on segmented step-response data, comparing four candidate model structures derived from McRuer’s theory. The results identified a second-order model with a derivative state and a reaction delay component as the optimal structure for approximating driver behavior in this specific task. This model requires only four parameters, which are jointly determined by the driver’s mental state and the testing conditions. The study demonstrated that increasing model complexity did not improve prediction accuracy for this linearizable task. Furthermore, the large dataset of 92 drivers enabled the calculation of confidence intervals for these parameters, a statistical advantage over previous studies that typically involved only 4 to 10 participants. The findings were verified by establishing a relationship between the selected single-loop model and more complex multi-loop models found in recent literature. The significance of this work lies in providing a parsimonious, interpretable mathematical description of driver dynamics that facilitates easier comparison between individuals and reduces errors associated with parameter uncertainty. By validating that a simple linear model suffices for well-defined lane-changing tasks, the study supports the use of such models in the design of assistance systems and shared control architectures. The approach offers a robust method for evaluating driver performance and mental state through dynamical parameters, contributing to the broader field of transport safety and human-machine interaction.

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 2026-06-25; verification: verified.

Topics

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

• lane changing
• situational awareness
• mental model of traffic
• lane positioning
• traffic density
• simulator validity fidelity

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).

• Empirical Findings: behavioral performance data
• Theoretical Contribution: computational model, theory or model