Optimization of Multi-Source Data Fusion Algorithm in Environmental Perception of Intelligent Connected Vehicles

Background: Achieving high-precision, real-time environmental perception in complex traffic environments is critical for intelligent connected vehicles, but existing multisource data fusion methods face spatiotemporal alignment, confidence assessment, and efficiency issues. Methods: A fusion algorithm combining improved EKF and deep learning is proposed: CNN extracts image semantics, point cloud segmentation processes 3D info; adaptive weighting dynamically evaluates sensor confidence; EKF achieves spatiotemporal synchronization; Bayesian reasoning optimizes results. Results: The algorithm enhances detection accuracy (93.2 %), tracking precision (0.15 m), efficiency (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$39 ~\text{ms} /$</tex> frame), and robustness (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\pm 1 \%$</tex> fluctuation). Concluding Remarks: It solves key issues in traditional methods, providing an innovative solution for high-precision perception in complex environments.