Investigating the Effects of Cooperative Driving for CAVs in Different Driving Scenarios Using Multi-Driver Simulator Experiments

Yue, Lishengsa; Abdel-Aty, Mohamed; Wang, Zijin · 2022 · ROSA P / Safety Research Using Simulation (SAFER-SIM) 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 study addresses the gap in research regarding cooperative driving strategies (CDS) that account for human drivers and multi-agent decision-making in mixed traffic environments. While connected and automated vehicle (CAV) technology promises improved safety and efficiency, existing algorithms often assume fully autonomous control or ignore the complexities of human-machine interaction. The authors specifically target two high-conflict locations: non-signalized intersections and freeway off-ramps. The research aims to develop and evaluate CDS algorithms that function effectively with varying market penetration rates of connected vehicles (CVs) and CAVs, design effective human-machine interfaces (HMIs) for driver guidance, and apply multi-agent reinforcement learning for complex diverging maneuvers. The researchers employed a self-developed human-in-the-loop co-simulation platform integrating the CARLA driving simulator and the SUMO microscopic traffic simulator. The SUMO model was calibrated using real-world trajectory data extracted from drone videos recorded at a busy three-way stop intersection at the University of Central Florida. The study was divided into three tasks. Task 1 involved developing an efficiency-oriented CDS for non-signalized intersections and testing it across different automation levels. Task 2 focused on designing and evaluating three different HMIs for providing real-time speed guidance to CV drivers. Task 3 utilized a multi-agent deep-Q network (MADQN) to train a cooperative decision-making strategy for freeway off-ramp diverging scenarios. The results demonstrated significant improvements in traffic performance. For Task 1, the proposed CDS reduced travel time by up to 53.8% in mixed CV-human-driven vehicle environments, 66.4% in CV-CAV environments, and 73.7% in fully CAV environments. In Task 2, the evaluation of HMIs revealed that a graphic-based interface was superior for displaying minor speed change requirements, offering better precision and guidance for drivers approaching intersections compared to other designs. For Task 3, the trained MADQN model significantly outperformed baseline models in terms of efficiency and safety while maintaining a high successful diverging rate. The significance of this work lies in its practical approach to implementing cooperative driving in realistic, mixed-traffic conditions rather than idealized fully autonomous scenarios. By validating strategies through multi-driver simulator experiments, the study provides evidence that cooperative driving can substantially reduce travel time and improve safety even with partial market penetration. Furthermore, the findings on HMI design offer actionable insights for reducing cognitive load and improving driver acceptance of cooperative driving systems. The application of multi-agent reinforcement learning for off-ramp diverging also advances the state-of-the-art in handling complex, multi-vehicle interactions, suggesting that model-free methods can effectively manage the computational demands of heavy traffic circumstances.

Key finding

The proposed cooperative driving strategy reduced travel time by up to 73.7% in CAV environments, graphic-based HMIs improved driver guidance precision, and multi-agent reinforcement learning significantly enhanced off-ramp diverging efficiency and safety.

Methodology

simulator

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.

• in vehicle coaching
• traffic density

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
• Methodological Resource: tool software
• Theoretical Contribution: computational model