UAV target tracking: a survey

Abstract

Unmanned Aerial Vehicles (UAVs) have become critical enablers of integrated air-space-ground Internet of Things (IoT) ecosystems, with target tracking serving as a foundational technology. This paper classifies UAV target tracking into two distinct paradigms: active tracking and passive tracking, differentiated by their operational scopes and technical objectives. Active tracking is defined as a closed-loop spatial pursuit system, whereby UAVs dynamically track targets through iterative cycles centered on three primary stages: online passive tracking, state fusion estimation, and tracking strategy generation, with subsequent execution phases implied in the loop. This workflow bridges perception and action, enabling spatial engagement through continuous sensor-to-control feedback. In contrast, passive tracking acts as a vision-centric analytical module that exclusively extracts target image-domain attributes from visual sensors—devoid of physical state inference or control mechanisms. As a preprocessing stage for active systems, it is constrained to the visual perception layer, lacking the spatial engagement capabilities inherent in closed-loop tracking systems. This paper conducts an in-depth analysis of the application, key challenges, and future trends in both active and passive UAV target tracking. By systematically discussing the relationships among relevant technologies, this work aims to establish a foundational reference framework and offer citation material for guiding the future development of UAV target tracking technologies.