Long-distance drone classification in intracity environments via temporal single-photon detection

Abstract

Single-photon light detection and ranging (SP-LiDAR) offers exceptional capabilities for long-range imaging or target identification, yet its high system complexity and substantial power/time consumption have hindered deployment in task-oriented applications. In this paper, we present the first experimental demonstration of long-distance drone classification beyond 5 km in a real-world intracity setting using temporal single-photon LiDAR (TSP-LiDAR). Unlike conventional approaches, TSP-LiDAR directly extracts target features from temporal histograms, eliminating the need for imaging optics and computationally intensive reconstruction algorithms. Both simulation and field results confirm that our TSP-LiDAR successfully achieves high classification accuracy under challenging conditions, including low signal-to-noise ratio (SNR) and sparse echo photon count. Specifically, in our 5 km experiments, we achieve pose and type classification accuracies of 96.11% and 98.06%, respectively, at SNRs ranging from 0.01 to 0.15 and echo photon count rates between 2∼30 kHz. Notably, the system further discriminates fine pitch variations down to approximately 1.15°. These findings underscore the potential of our TSP-LiDAR for robust, long-range classification of small aerial targets, paving the way for advanced urban surveillance and air defense applications.