Adaptability analysis of autonomous vehicles on small-radius circular curves based on co-simulation

This study establishes a high-fidelity Prescan-CarSim-Simulink co-simulation framework to evaluate the adaptability of autonomous vehicles (AVs) on small-radius circular curves (60–700 m). The results reveal that when the critical adhesion coefficient (μ) exceeds 0.2 at curve radii ≤200 m, ride comfort degrades to "moderately uncomfortable" levels. However, speed-radius coupling (70–100 km/h with typical minimum radii) improves comfort to Levels 1–2. Current road standards fail to meet AVs' comfort needs at minimum radii, providing insights for road design optimization and AV system upgrades.

With the rapid development of autonomous driving technology, the incompatibility between traditional road infrastructure and autonomous vehicles (AVs) has become increasingly prominent. Current road alignment design standards, primarily based on human-driven vehicles, do not fully account for the unique operational characteristics of AVs. This mismatch is particularly evident in challenging scenarios like small-radius circular curves, where AVs often face issues with lateral stability control, path-tracking precision, and ride comfort. While multi-software co-simulation and digital twin technologies have emerged as key tools for AV testing, existing research has focused more on large-radius curves, leaving a gap in systematic analysis of small-radius circular curves—especially regarding quantitative evaluations of safety and comfort.

To address this gap, this study focuses on small-radius circular curves and systematically analyzes AVs' dynamic responses under varying speeds and curve radii (from minimum limit radii to general minimum radii). A high-fidelity Prescan-CarSim-Simulink co-simulation platform is employed, featuring a decoupled lateral-longitudinal control strategy: discrete Linear Quadratic Regulator (LQR) for lateral trajectory tracking and Proportional-Integral-Derivative (PID) control for longitudinal speed regulation, forming a closed-loop autonomous driving system. Test scenarios, constructed following Germany's PEGASUS project framework (functional-logical-concrete hierarchy), include 400-m tangent segments (speed stabilization zones), 250-m transition curves (smooth curvature transitions), and circular curves with radii referenced to China's Design Specification for Highway Alignment (JTG D20-2017). The Audi A8 Sedan is selected as the test vehicle, with a 6-degree-of-freedom CarSim model and 1000-Hz sampling frequency to capture transient dynamics. Key evaluation metrics include speed (40–100 km/h), lateral acceleration (Ay), and critical adhesion coefficient (μ), with comfort levels classified into four grades based on German research (Level 1: comfortable; Level 4: uncomfortable).

Simulation results show that under low-speed conditions (40–60 km/h) with radii approaching the minimum limit (R ≤200 m), peak lateral acceleration (0.2–0.4g) and critical adhesion coefficient (0.15–0.22) significantly reduce safety margins and degrade comfort to Level 3 (moderately uncomfortable). In contrast, higher speeds (70–100 km/h) combined with larger radii (general minimum) stabilize dynamic responses, improving comfort to Levels 1–2. Notably, current road standards (lateral friction coefficients 0.10–0.17) meet conventional vehicles' safety needs but not AVs' comfort requirements at minimum radii, where μ often exceeds 0.20. These findings are validated by comparisons with Wang et al.’s simulations, confirming that low-speed, tight-curve conditions (R ≤200 m, 40–60 km/h) harm comfort, while appropriate speed-radius combinations mitigate adverse effects.

This study provides a theoretical basis for road alignment design in autonomous driving environments and practical recommendations for retrofitting existing roads and optimizing AV systems. Future work will focus on validating AV performance under low-friction conditions (e.g., rain, snow), simulating multi-vehicle interactions, optimizing sensor fusion algorithms for complex environments, and exploring V2I-based predictive control—ultimately refining AVs' Operational Design Domains (ODD) and supporting intelligent infrastructure upgrades.

This paper was published in Smart Construction. Smart Construction is a peer-reviewed, open-access journal focused on publishing original research works, communications, reviews, perspectives, reports, and commentaries in all areas of intelligent construction, operation, and maintenance, covering both fundamental research and engineering applications. The journal aims to be a top-tier journal in this field with targeted Impact Factor of 8. Currently, the APC is fully waived for the author.

Citation: Zhang Z, Yu X, Zhu L, Wang M. Adaptability analysis of autonomous vehicles on small-radius circular curves based on co-simulation. Smart Constr. 2025(3):0018, https://doi.org/10.55092/sc20250018.