ABSTRACT Smart irrigation systems are used to control, monitor and automatically irrigate fields based on artificial intelligence techniques. Irrigation decision‐making is the core content of smart irrigation systems, which aim to guide efficient irrigation and ensure the sustainable use of water. Traditional irrigation decision‐making methods have encountered difficulties in accessing meteorological data and high sensor costs. Machine learning (ML) models and multisource irrigation strategies are expected to provide more precise and cost‐effective irrigation management. This paper provides an overview of the application of traditional irrigation decision‐making methods in smart irrigation systems and identifies ML models as the key to the transformation from traditional agriculture to smart agriculture. ML models upgrade irrigation management from experience‐based to data‐driven by capturing complex environmental‐crop relationships to optimize water use. However, they face issues such as poor model interpretability, data shortages and barriers to popularization in small‐scale farms. Future work should focus on developing closed‐loop systems by integrating emerging technologies and combining ML models with edge computing, strengthening data privacy protection, and simplifying operational interfaces to unlock their full potential for precise and sustainable irrigation.
Liu et al. (Mon,) studied this question.