Assessing the potential of integrating automation and artificial intelligence across sample-destructive methods to determine plant water status: A review and score-based evaluation

• Published in: Computers and electronics in agriculture Vol. 224; p. 108992
• Main Authors: Mucchiani, Caio, Zaccaria, Daniele, Karydis, Konstantinos
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.09.2024
• Subjects: Agricultural automation; Agricultural robotics; Artificial intelligence; Plant water stress; Resource-efficient water management; Plant water stress; Agricultural robotics; Agricultural automation; Artificial intelligence; Resource-efficient water management
• ISSN: 0168-1699; 1872-7107
• DOI: 10.1016/j.compag.2024.108992

Sample-destructive methods for the determination of plant water status have been the primary reference for various agronomic practices over the years. Several recent technological advancements in automation, robotics, and artificial intelligence (AI) have helped make progress toward more resource-efficient water management. However, several methods, especially those conducted in situ, still require considerable labor and can be further improved via the integration of automation. To this end, this review article has a twofold aim. (1) To point out relevant aspects and technological considerations of sample-destructive methods for determination of plant water status in comparison to proximal and remote monitoring technologies, while also illustrating interrelations among the different measurement practices. (2) To evaluate the potential of current methods to be automated and endowed with AI capabilities that can further enhance the methods’ outcomes such as accuracy, precision, and consistency. To address the first objective, 97 articles were downselected and included in a meta-analysis performed in this review article from an initial literature survey comprising 550 articles related to the determination of plant water status over a ten-year time frame. The methods developed and reported within the selected articles were classified based on several key features such as type of measurements, required equipment, sampling time, location of measurements, need for calibration, and affordability. To achieve the second aim, an automation score based on several key metrics was proposed and then used to rank the different methods in terms of potential for automation. This work can spark further discussions within the agricultural engineering community at a time when automation and AI efforts in agriculture create new challenges and opportunities for improving the determination of plant water status in support of more resource-efficient agricultural water management. •Review of different methods to determine plant water status.•Focus on several sample-destructive methods and presentation of common and distinctive features for each method.•Propose an automation score and evaluate existing methods based on such score to determine their potential toward automation and artificial intelligence integration.