Review: unmanned aerial vehicles and artificial intelligence in precision agriculture

B. Zvara*, M. Macák and J. Galambošová

Slovak University of Agriculture, Faculty of Engineering, Department of Machines and Production Biosystems, Institute of Agricultural Engineering, Transport and Bioenergetics, Tr. Andreja Hlinku 2, SK949 76, Nitra, Slovakia
*Correspondence: benjaminzvara@gmail.com

Abstract:

To meet the needs of sustainable intensification in crop and animal production, farmers use a set of technologies which are referred to as Agriculture 4.0 to 5.0 or digital agriculture. Differences compared to traditional precision farming techniques are in extensive use of UAV, smart sensors implemented in machines, crops, animals and in the soil, cloud computing, IoT, together with extensive use of AI for data analyses. Unmanned Aerial Vehicles (UAV), also called drones, have become an essential tool in digital agriculture. UAVs have witnessed remarkable development in the past decades and so in the recent years, the topic of agricultural UAVs has gained the attention of many farmers. The submitted paper provides a review on recent scientific literature dedicated to the utilization of agricultural UAVs. The utilization areas are reviewed in monitoring (remote sensing), interventional applications of various inputs, and other areas of possible utilization. The novelty of this review highlights the importance of the integration of UAVs with artificial intelligence (AI) and the Internet of Things (IoT). Sophisticated artificial intelligence and machine-learning algorithms are developing to analyse UAV-collected data, enhancing the accuracy and efficiency. Machine learning models in combination with artificial intelligence are capable of yield prediction and crop management, effecting future decision-making processes. Several key opportunities can be identified for future research, including the development of more sophisticated decision-making processes and machine learning methods based on artificial intelligence, the automation of agricultural crop production, improved UAV autonomy, and the potential use of UAV swarms in different field operations.

Key words:

artificial intelligence, drone; precision agriculture, unmanned aerial spraying system, Unmanned aerial vehicle

Digital elevation models as predictors of yield: Comparison of an UAV and other elevation data sources

D. Moravec¹*, J. Komárek¹, J. Kumhálová², M. Kroulík³, J. Prošek¹ and P. Klápště¹

¹Czech University of Life Sciences Prague, Faculty of Environmental Sciences, Department of Applied Geoinformatics and Spatial Planning, Kamýcká 129, CZ165 21 Prague, Czech Republic
²Czech University of Life Sciences Prague, Faculty of Engineering, Department of Machinery Utilization, Kamýcká 129, CZ165 21 Prague, Czech Republic
³Czech University of Life Sciences Prague, Faculty of Engineering, Department of Agricultural Machines, Kamýcká 129, CZ165 21 Prague, Czech Republic
*Correspondence: dmoravec@fzp.czu.cz

Abstract:

Topography usually plays an important role for yield variability assessment. This study provides insight into the use of surface models from different sources for agriculture purposes: unmanned aerial vehicle imagery, LiDAR data and elevation data acquired from a harvester. The dataset from an aerial vehicle was obtained in the form of ortho-mosaics and digital surface model using casual camera. The LiDAR data was provided by the State Administration of Land Surveying and Cadastre in the form of Digital Terrain Model of the 4th and 5th generation. The data of yield together with its coordinates were gained from a combine harvester in the form of a regular grid. Yield data was interpolated by kriging geostatistical method. Position data including an altitude was used for modelling the last digital surface model. All gained surface models were correlated with the spring barley yield. Results show correlation similarity across all tested models with the yield; no significant differences were sighted. Free available coarser scale data is able to predict a yield sufficiently. The study indicates less effectivity of using very detailed scale data sources due to its time-consumption or expensive data gathering and processing process.

Key words: