基于信息融合的智能网联汽车安全交互决策

In open traffic scenarios, intelligent connected vehicles still face critical bottlenecks such as weak safety and reliability, and insufficient interactive attributes. With the advancement of artificial intelligence (AI) and breakthroughs in deep learning, AI models have made significant advancements in the field of autonomous driving, applicable to scene understanding and reasoning in autonomous driving. This paper provides a comprehensive review of the research on safety interactive decision-making for intelligent connected vehicles based on information fusion. It begins by organizing research on traffic perception and understanding in open scenarios, then explores decision-making and planning models with social interaction attributes, and concludes with an examination of safety verification techniques for AI model hallucinations. By integrating research in these three areas, fully leveraging the powerful capabilities of AI models to achieve the driving skills of “skilled human drivers”, and discussing safety assurance technologies to compensate for the “occasional mistakes” of AI models, it is hoped to address the long-tail safety issues in autonomous driving and further advance the development of autonomous driving technology.