Multimodal Perception and Generative Reality for Autonomous Driving

The work presented in this book was supported by Jiangsu University of Science and Technology. Autonomous driving algorithms, as a pivotal direction in intelligent transportation, are fundamentally transforming the future of mobility. In recent years, fueled by advances in deep learning and high-performance computing platforms, autonomous driving systems have achieved substantial progress in structured environments and are steadily transitioning towards deployment on open roads. Among key enablers, multimodal perception has emerged as a core capability, integrating complementary data from sensors such as LiDARs, cameras, and radars to enhance environmental awareness, semantic understanding, and behavior prediction. Particularly in complex traffic scenarios, the synergy across different modalities is essential to improving perception accuracy, robustness, and decision-making safety.