Research on multi-source information fusion perception and path optimization algorithm of new energy vehicle-intelligent connected vehicle

In this paper, the multi-source information fusion perception and path optimization algorithm of new energy vehicle-intelligent connected vehicle (NEV-ICV) is studied. NEV-ICV aims to achieve a safer, more efficient and low-carbon travel experience by integrating multi-sensor perception, vehicle networking communication and AI algorithm. However, the existing technologies still face challenges in the aspects of heterogeneous data fusion, target detection accuracy and collaborative optimization of energy consumption and efficiency of path optimization in complex environment. In this paper, a cross-modal attention fusion framework based on Transformer is proposed, which is used to realize efficient fusion of multi-source heterogeneous sensor data and improve the perception accuracy and robustness in complex traffic environment. At the same time, a two-level dynamic path optimization architecture is designed, which combines the improved A* algorithm with deep reinforcement learning to realize global and local collaborative multi-objective path planning. The results of simulation and real vehicle test show that the proposed method has obvious advantages in energy consumption, real-time performance and safety, and provides a solid technical support for mass production of NEV-ICV.