Abstract: Precision agricultural aviation enables large-scale, high-efficiency farmland information acquisition and precision farming through aerial platforms. Its core principle is achieving on-demand variable-rate operations based on farmland spatiotemporal heterogeneity, optimizing resource allocation and protecting ecology while ensuring crop yield. With global population growth and resource constraints, this technology has become strategic for food security and sustainable agriculture. Unmanned aerial vehicles have driven innovation across precision agricultural aviation components. At perception level, multispectral, hyperspectral, and LiDAR sensors enable multi-dimensional inversion of crop nutrition, water, pests, and structure. At decision-making level, variable-rate prescription models and path planning algorithms have been optimized. At execution level, advances in precision flight control and variable-rate devices have been achieved. Despite these accomplishments, research predominantly focuses on single components, with few studies addressing systematic design and collaborative optimization of perception, decision-making, and execution as an integrated whole. Information transmission delays, inability of decision models to respond real-time to perceptual changes, and lack of effective feedback from execution constrain system performance, preventing formation of a true technological closed loop from data acquisition to precision operation. The construction of an integrated perception-decision-execution closed-loop system aims to achieve organic collaboration among the three core components. Intelligent perception relies on multi-source sensor fusion technologies to realize real-time acquisition and on-board analysis of multi-modal farmland data. Intelligent decision-making integrates agronomic mechanisms with optimization algorithms to generate path planning schemes and variable-rate operation prescriptions adapted to spatial heterogeneity. Precision execution ensures reliable implementation of decision instructions through high-precision flight control and variable-rate actuation mechanisms, while feeding back operational outcomes to the perception layer for subsequent corrections. This closed-loop paradigm overcomes the limitations of traditional open-loop operations and propels precision agricultural aviation from automation toward autonomy. This paper provides a systematic review of the perception-decision-execution system in precision agricultural aviation. First, it examines the current research status of key technologies across the three core components. In intelligent perception, the review analyzes multi-source sensing technologies including multispectral, hyperspectral, LiDAR, and SAR, along with the challenges of multi-modal data fusion. It also discusses embedded AI and real-time on-board processing architectures, focusing on end-edge-cloud collaborative frameworks and lightweight visual perception models such as YOLO and MobileNet. In intelligent decision-making, the paper outlines the evolutionary path from reactive control to optimized planning and learning-based adaptation. It surveys intelligent path planning methods for static and dynamic environments and evaluates variable-rate decision models for fertilization, spraying, and seeding, highlighting bottlenecks related to real-time performance and generalization capabilities. In precision execution, the review summarizes intelligent flight control methods for trajectory tracking and disturbance rejection, and examines the technical characteristics of spray systems including flow control and electrostatic spraying, as well as granular material spreading devices including centrifugal and pneumatic types. Second, this paper systematically integrates cross-disciplinary research achievements including multi-source sensor fusion, embedded processing, path planning, variable-rate decision-making, flight control, and actuator design, constructing a comprehensive technical knowledge graph for the field. Finally, it analyzes the multiple challenges facing the transition toward large-scale applications, including perception robustness, real-time decision-making capability, execution accuracy, and system integration. It also discusses critical issues in the collaborative optimization and validation of perception-decision-execution systems, noting the current lack of unified interface standards and systematic evaluation protocols across different crops, operations, and environmental conditions. Looking forward, the development of precision agricultural aviation closed-loop systems is expected to move beyond linear improvements in individual technologies toward deeper integration with emerging information technologies. Future research directions include embodied intelligence with continuous environmental interaction learning, enabling UAVs to autonomously acquire skills through physical interaction with farmland environments; air-space-ground integrated swarm intelligence networks, where heterogeneous UAV clusters collaborate with satellite remote sensing and ground-based IoT sensors; scenario-specific algorithm-hardware-chip co-design, developing dedicated AI chips for agricultural edge computing; and human-machine collaborative hybrid-augmented intelligence systems, where AI handles data processing and precision operations while human experts oversee strategic planning and anomaly handling. By providing a systematic review of the current state, challenges, and future trends, this paper aims to offer researchers a valuable reference for understanding the technological development and identifying future research directions in this interdisciplinary field.