Deep Learning-Based Robust Visible Light Positioning for High-Speed Vehicles

Robustness is a key factor for real-time positioning and navigation, especially high-speed vehicles. While visible light (VLP) based on LED illumination image sensors widely studied, most of the VLP systems still suffer from high latency blurs caused by movements. In this paper, robust system vehicles proposed deep learning data-driven approach. The can significantly increase success rate decoding VLP-LED user identifications (UID) blurred images reduce computational detecting extracting stripe regions captured images. Experimental results show that UID using BN-CNN model could be higher than 98% when traditional Zbar-based decoder falls to 0, while time decreased 9.19 ms supported moving speed our scheme achieve 38.5 km/h. Therefore, enhance robustness against movement guarantee response navigation