Intersection Complexity and Its Influence on Human Drivers

Weinreuter, Hannes; Strelau, Nadine-Rebecca; Qiu, Kevin; Jiang, Yancheng; Deml, Barbara; Heizmann, Michael · 2022 · DOAJ

DOI: 10.1109/ACCESS.2022.3189017

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 investigates how intersection complexity influences human driving behavior, a critical factor for developing automated vehicles capable of navigating mixed traffic in urban environments. The research focuses specifically on unsignalized intersections, which lack traffic lights or priority signs and rely on the "right-before-left" rule, creating ambiguous situations that challenge both human and automated drivers. The authors aim to define intersection complexity through a comprehensive set of static and dynamic features and determine if these features can predict typical human driving behaviors, such as decision-making distances and speed adjustments. To achieve this, the researchers conducted a naturalistic driving field study in Karlsruhe, Germany, involving 34 participants who drove a sensor-equipped test vehicle through 17 unsignalized intersections (13 T-junctions and 4 X-intersections). The study utilized a 16-channel LiDAR sensor, IMU, and GPS to record trajectories and environmental data. The authors defined intersection complexity using static features (street width, visibility distance, number of trees) and dynamic features (number of vehicles and pedestrians, priority relationships, entry location, and turning direction). Driver behavior was quantified using three trajectory-derived metrics: commit distance (the distance at which a driver commits to proceeding), velocity drop during the approach, and minimal velocity. Data processing involved SLAM for precise tracking, lanelet mapping for intersection geometry, and clustering algorithms to detect and classify surrounding traffic participants. The results demonstrate that entry location and turning direction significantly affect driver behavior features. Furthermore, the study confirms that human driving behavior at these intersections can be predicted using the defined complexity features. The authors also explored dimensionality reduction techniques, testing reduced feature sets and autoencoders, and found that prediction remains feasible with condensed representations of intersection complexity. This indicates that a robust model of intersection complexity can effectively capture the essential factors influencing driver decisions without requiring the full set of raw environmental data. The significance of this work lies in its contribution to the development of automated vehicles, particularly for handling ambiguous, unsignalized intersections common in European cities. By establishing a predictive model of human behavior based on intersection complexity, automated systems can anticipate likely human actions, leading to safer and more reliable decision-making in mixed traffic scenarios. Additionally, the findings offer insights for intersection design, potentially helping to reduce collision risks by understanding how specific environmental features influence driver perception and reaction. The study provides a detailed framework for quantifying intersection complexity that integrates both static infrastructure and dynamic traffic conditions, addressing a gap in previous research that often considered only limited subsets of these factors.

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.

• traffic density
• driver vru interaction