Research on Machine Learning Models for Sensor Fusion of Intelligent Connected Vehicles

This paper proposes a sensor fusion method based on the support vector machine (SVM) algorithm, which aims to improve the environmental perception ability of intelligent connected vehicles by fusing data from different sensors. Along with the fast development of ITS technology, the high efficiency and precise environment sensing system is very important in the autonomous driving. As a powerful classification algorithm, SVM is particularly suitable for data processing in high-dimensional feature space and can effectively classify and analyze multimodal sensor data. This paper discusses in detail the application of SVM in multimodal data processing, especially in the complex sensor data fusion process, how SVM can effectively divide different target categories by constructing high-dimensional hyperplanes, thereby improving the accuracy of target recognition and detection. Experimental results show that the SVM algorithm shows excellent accuracy and robustness in target detection tasks under different weather and lighting conditions, especially in complex environments (such as night and rainy days). Specifically, in complex environments, the accuracy of the SVM algorithm is improved by more than 12%, and the recall rate reaches more than 95%, which is significantly better than traditional Kalman filtering and simple deep learning methods.