Ultra-Wideband Photoelectric Fusion Beam Management and Multi-Node Cooperative Perception for Vehicle Networking

Abstract

This study concentrates on the perception instability caused by the lack of multi-node information due to the fluctuation of communication links in the Cooperative Perception (CP) of Vehicle-to-Everything (V2X) on real roads. To alleviate this critical bottleneck and improve the continuity and cooperation efficiency of multi-vehicle environment understanding, an Ultra-Wideband Optical-Photonic Beam Management Cooperative Perception Network (UWB-OPBM-CPNet) model is proposed, which integrates CP and communication enhancement capabilities. The model takes Vehicle-to-Vehicle Transformer++ as the backbone and adds dynamic topology modelling, cross-node compensation and a lightweight timing cache to stabilise the cooperative characteristics. At the same time, fast and robust beam direction finding is realised by using photoelectric direction prior and UWB pulse time resolution and the link quality and cooperative stability are improved from the communication side. Based on V2X Simulation (V2X-SIMn), Raymobtime and Dataset for Autonomous Driving and Intelligent Roads-V2X (Dair-V2X), the verification shows that the model still obtains 63.78% 3D-mean Average Precision under the condition of serious node packet loss; it also maintains 70.88% accuracy under the condition of asynchronous time series, both of which are better than the baseline. On the communication side, the photoelectric-UWB fusion shortens the beam search time to 12.35 ms and achieves a link survival rate of 88.76% in the heavy occlusion environment, which markedly enhances the stability of data transmission. The results confirm that UWB-OPBM-CPNet can improve the cross-node CP and communication performance in a complex road environment. It can provide a stable cooperation scheme with engineering value for intelligent networked vehicles, road infrastructure cooperation systems and vehicle-road integration applications.