Autonomous Vehicles and Transportation Safety: A Comprehensive Review of Risk and Challenges

Engineering notes

Key topics: autonomous driving, autonomous vehicle, lidar, deployment. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Autonomous vehicles (AVs) are regarded as a cornerstone of future intelligent transportation systems, yet their safe deployment remains a critical challenge. This review provides a systematic analysis of the safety risks and challenges of AVs across three domains: road traffic safety, functional safety, and cybersecurity. Using the PRISMA framework, 62 peer-reviewed studies published between 2016 and 2024 were analyzed and synthesized. The findings reveal that while AVs have the potential to reduce human-related crashes and improve mobility, the current literature remains fragmented. In road traffic safety, predictive crash models and validation frameworks show promise but remain limited in addressing rare and unpredictable scenarios. In functional safety, traditional methods such as hazard analysis and failure mode assessment struggle to account for AI-driven uncertainties, while emerging approaches like SOTIF and simulation-based testing lack standardized implementation. In cybersecurity, vulnerabilities such as GPS spoofing, LiDAR manipulation, and network intrusions continue to pose significant threats, and current defenses remain largely reactive. This review highlights that AVs safety challenges are systemic and interconnected, requiring integrated approaches that cut across technical, regulatory, and human dimensions. Key future directions include hybrid validation ecosystems combining real-world and simulated testing, explainable and trustworthy AI models, proactive cybersecurity architectures, and harmonized international standards. By synthesizing current knowledge and identifying research gaps, this study provides a foundation for advancing safe, secure, and socially accepted AVs deployment. This review’s novelty lies in its cross-domain synthesis: rather than treating road traffic safety, functional safety, and cybersecurity separately, it analyzes how failures propagate across these domains and identifies where existing standards and evaluation practices fall short for AI-driven autonomy.

Links and sources

• PDF from original source

Comments