Impact Factor:9.9

DOI number:10.1016/j.aei.2025.104037

Journal:Advanced Engineering Informatics

Key Words:Lane-free roundaboutAutonomous vehiclesSelf-organizingMotion planningIntelligent agent model

Abstract:As autonomous driving advances, Lane-Free Roundabouts (LFRs) offer a promising alternative to traditional lane-based roundabout designs, presenting the potential to significantly improve traffic efficiency. However, enabling non-connected Autonomous Vehicles (AVs), which rely solely on onboard sensors, to independently execute key maneuvers (entry, circulation, and exit) while maintaining smooth traffic flow in LFRs remains a critical challenge. To address this gap, this study proposes a self-organizing motion planning framework (LFR-MPF) for non-connected AVs in LFRs. The framework integrates an enhanced Intelligent Agent Model (IAM) to simulate microscopic vehicle interactions, a Target Guidance Model for flexible path adjustment, and a self-organizing yielding strategy based on local sensing. Through these components, LFR-MPF enables AVs to safely navigate lane-free environments while maximizing road space utilization. The LFR-MPF model demonstrated both human-like behavioral realism and superior traffic efficiency. Calibration and initial validation on the real-world RounD dataset confirmed its human-like driving patterns. Subsequent high-density simulations showed LFR-MPF boosted capacity at the Place Charles de Gaulle roundabout to ∼ 6000 veh/h, substantially exceeding both advanced MPC-based AVs (∼4700 veh/h) and human-driver baselines (∼3600 veh/h) while minimizing delays. Finally, sensitivity analyses affirmed the framework’s robustness. LFR-MPF offers a practical and effective motion planning solution, demonstrating a viable path toward deploying efficient and safe non-connected AVs in future unstructured traffic environments.

Note:https://doi.org/10.1016/j.aei.2025.104037

Indexed by:Article

Translation or Not:no

Date of Publication:2025-11-02

Included Journals:SCI

Links to published journals:https://doi.org/10.1016/j.aei.2025.104037