Artificial Intelligence Systems for Solving Problems of Agro­Industrial Complex Digitalization and Robotization

Abstract

The digital transformation of agricultural technological processes is substantiated to be relevant in the context of the exacerbating global problems of food security, the agricultural business stagnation in the regions of the northern Non-Black Earth Region, as well as the ESG transformation of the economy.Research purpose To develop control algorithms for unmanned aerial vehicles (UAVs) based on numerical methods of machine learning to ensure the monitoring of the crops state and the improving of the production process planning and operational management.Materials and methods The following methodology was used: the original methods of machine learning, knowledge engineering and computer modeling for organizational and technological processes of technical objects’ life cycle in industry and products’ life cycle in the national economy, as well as mathematical and algorithmic models, methods and prototypes of proactive automation tools for information, physical and energy interaction of heterogeneous robotic and cyber-physical complexes.Results and discussion Artificial intelligence systems were created for the photogrammetric processing of visible spectrum images and those taken with multispectral video cameras with the construction of orthophotomaps, digital elevation models. Machine learning numerical methods were applied. Possible ways of formulating recommendations for the land revegetation and amelioration were demonstrated. Algorithmic software and hardware have been developed for the automation of vertical farms, closed cycle fish farming plants. The authors carried out wireless registration of measured and calculated parameters received from the distributed sensors, conducted their analysis based on big data technologies and proactive control of cyber-physical devices responsible for the functioning of the aqua and phytocultures life support systems. The authors provided the examples of produced UAVs and attachments designed for processing the agricultural land, as well as examples of automation modules for vertical farms that provide proactive autonomous control.Conclusions It was determined that the developed software and hardware ensured a 6-percent increase in the residual charge of the UAV battery after the flight. Image analysis using a multispectral camera improved the accuracy of identifying the plant areas with phytopathologies up to 99 percent.