Answers are grounded in this paper and cite their sources.
Monitoring extensive grazing systems is challenging because livestock are spread across large, diverse areas, limiting continuous observation. Although Artificial Intelligence (AI) and Unmanned Aerial Vehicle (UAV) technologies have significantly advanced livestock detection, tracking, behavioral analysis, and lameness assessment, most existing approaches remain perception-focused and primarily reactive in nature. This review pursues two main objectives. First, it systematically maps AI-based livestock movement prediction methods with an emphasis on UAV trajectory optimization for health monitoring. Second, it evaluates the degree of integration among livestock movement prediction, UAV sensing, trajectory control, and health-aware decision-making, and identifies the structural barriers to predictive monitoring architectures. Following the PRISMA 2020 guidelines, 57 studies from 2015 to 2026 were analyzed across sensing, analytics, forecasting, and UAV operations. The reviewed studies showed strong maturity in perception-oriented tasks, particularly behavior analysis and lameness assessment, which accounted for 61.4% of the reviewed studies. In contrast, UAV-based monitoring was reported in 19.3% of the studies, movement forecasting in only 3.5%, and no study (0%) demonstrated end-to-end closed-loop integration of sensing, forecasting, and UAV control. The main limitations included small longitudinal datasets, inconsistent validation, fragmented sensing–prediction–control workflows, and poor cross-layer interoperability. To address these gaps, a closed-loop UAV architecture that combines multimodal sensing, temporal forecasting, adaptive UAV coordination, and welfare-aware monitoring is proposed. The main constraint is not the sensing capability but the absence of coordinated forecasting-driven operational systems essential for anticipatory and scalable livestock management.
No takes yet. Share an insight, caveat, or question.
Pandey et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: