Autonomous vehicles (AVs) face significant threats to their safe operation in complex traffic environments. Adver-sarial policy for scenario generation has been established as a robust paradigm for enhancing AV resilience against adver-sarial perturbations through proactive exposure to synthetically engineered safety-critical scenarios. Training an attacker within an adversarial policy, allowing the target AV to expose vulnerabilities through interaction with this attacker. However, adversarial policies in existing methodologies often get stuck in a loop of over-exploiting established vulnerabilities, resulting in poor exploration for AVs. To overcome the limitations, we introduce a pioneering framework termed the vulnerability-aware and curiosity-driven adversarial reinforcement learning policy. Specifically, during the traffic vehicle attacker training phase, a surrogate network is employed to fit the value function of the AV victim, providing dense information about the victim's inherent vulnerabilities. Subsequently, random network distillation is used to characterize the novelty of the scenario, constructing an intrinsic reward to guide the attacker in exploring unexplored territories. Experimental results demonstrated that the adversarial policy embedded within the attacker exhibited robustness in convergence and significantly enhanced the activation of policy exposure in learning-based AVs, outperforming both other adversarial modalities and alternative reinforcement learning approaches, with a notable reduction in crash rates. The code is available at https://github.com/caixxuan/VCAT.

Existing autonomous driving systems are optimized for egocentric efficiency metrics, which are in fundamental conflict with the socialized expectations and habitual patterns of human drivers. This contradiction stems from the traditional approach's dual neglect of the trade-offs between self and other in driving decisions, and the culturally rooted social qualities of traffic interactions. To this end, this paper proposes a dual-adaptation framework that integrates social value orientation (SVO) and human-in-the-Ioop(HITL) guidance, modeling vehicular interactions as competitive-cooperative agents by quantifying the social utility function, and dynamically calibrating the SVO parameters with the help of real-time human feedback. The method innovatively transforms abstract social preferences into mathematically tractable decision boundaries, enabling the human-vehicle co-evolutionary mechanism to contextualize self-adaptation according to the regional driving etiquette, and thus cracking the inherent contradiction between individual trajectory optimization and group traffic harmony. Empirical studies based on Highway Env driving scenarios show that compared with pure reinforcement learning methods, this method reduces human-vehicle interaction conflicts while maintaining self-vehicle efficiency. The research results provide a quantifiable interaction paradigm and a verifiable training architecture for the construction of culturally-aware autonomous driving systems through the deep coupling of computational social value modeling and human social intelligence.

In this study, we introduce an AR-based Meta-Vehicle Road Collaboration Testing System (AR-MVRTs), a significant advancement in autonomous driving testing. This system utilizes vehicle-road collaboration, Augmented Reality (AR), and Metaverse technologies for high-risk scenario simulation and dynamic interaction between virtual data and actual vehicles and infrastructure, enhancing verification and optimization methods. The core contribution is the innovative framework and component model, integrating AR with vehicle-road collaboration technologies for a comprehensive environment that includes real autonomous vehicles, virtual scenarios, and parameterized test settings. The AR-MVRTs method efficiently generates critical testing scenarios, significantly improving testing efficiency. An implementation case shows the system’s practical application, where AR-MVRTs demonstrated a 658-fold efficiency increase, completing tests in 8.5 hours compared to 5580 hours required by traditional test matrices. This integration accelerates the transition from theoretical research to practical applications and offers deep insights into autonomous vehicle safety and reliability. This research promises to advance autonomous driving technology and lay a solid foundation for its future development.

Traffic modeling is vitally important to the simulation test of autonomous driving, as on-road testing is not only tedious, inefficient, costly, and often unsafe. The fidelity of the traffic model is thus the key to achieving effective and efficient simulation test results. However, high-fidelity traffic modeling remains a challenge due to the complexity of vehicle movement and the dynamic spatiotemporal interactions among vehicles. In this paper, we propose a novel system-based traffic modeling approach that considers traffic as a whole, or a collection of all vehicles’ movement and their interactions involved simultaneously. In addition, a long short-term memory (LSTM) encoding-decoding framework with a multi-head self-attention mechanism is proposed to capture the temporal dependency of vehicle motion and represent the vehicle-vehicle interaction. Furthermore, some traffic data are collected and used for model training purposes. A metric representing one of the traffic attributes, as the criterion for data screening, is proposed to describe the traffic complexity and disorder, or chaos. Experimental results demonstrate that the proposed model has achieved better performance compared with others from the literature.