#软件架构

Year-round utilization of agricultural robotic platforms requires adaptable systems capable of performing multiple operations with interchangeable end-effectors. Although mechanical tool-changing mechanisms are well established, software interoperability, communication, and adaptive perception remain challenges. This narrative review examines software architectures, communication protocols, and edge intelligence technologies relevant to modular agricultural end-effectors. Publications identified through Scopus and Web of Science, complemented by citation exploration, are synthesized across five areas: hardware abstraction and middleware integration, deterministic fieldbuses, distributed network communication, edge computing and artificial intelligence (AI) deployment, and visual servoing. The literature highlights the need to separate task coordination from actuator control and to select communication protocols according to timing and bandwidth requirements. Edge computing, model optimization, and modular perception services can support AI deployment on resource-constrained robotic platforms. However, existing approaches primarily address individual integration challenges rather than complete software reconfiguration following tool exchange. Based on these findings, this review proposes the Modular Agricultural End-Effector Software Architecture (MAESA), a conceptual reference framework integrating automated tool identification, hardware abstraction, communication, perception, and control. Its theoretical contribution lies in unifying previously fragmented integration approaches, while its practical significance is providing design guidance for autonomous tool exchange and year-round agricultural operations.