Research on fused estimates event-based vision for object detection in high-speed Maglev trains

Robust 3D object detection remains a critical challenge in engineering perception systems for mid-speed magnetic levitation trains, where high velocity, real-time processing, and environmental variability must be simultaneously addressed. Existing LiDAR-based solutions suffer in adverse weather, Radar lacks spatial detail, and frame-based cameras introduce latency under dynamic conditions. To address these limitations, we propose a novel sensor fusion framework that integrates 4D LiDAR, 3D Radar, and event-based vision to improve object detection accuracy and temporal responsiveness. The system employs a confidence-weighted fusion mechanism that dynamically adapts to changing sensor reliability, enabling robust performance in poor visibility and high-speed transit scenarios. Experimental results demonstrate smoother, more reliable object tracking across varied conditions, showing advantages over individual sensor modalities. This research contributes a practical and scalable solution for enhancing autonomous perception in Maglev rail systems, supporting safer and more reliable operations in real-world transportation environments.