The integration of sensor technology and IoT in precision agriculture has significantly improved farm operations. These technologies enable farmers to monitor crop conditions, soil moisture, and equipment performance in real-time, leading to more efficient practices. However, challenges such as missing data due to low network connectivity in farming areas and inefficiencies due to resource overuse during field operations still persist. This paper addresses these critical issues by presenting and implementing a comprehensive data collection mechanism using IoT architecture and sensors installed on farm machinery, capturing key metrics in real-time using serverless cloud technology. This data helps optimize field coverage and minimize redundant work. We further address the issue of missing data incorporating artificial intelligence-based Bidirectional Recurrent Imputation for Time Series (BRITS) model. When considering random missing data, the model successfully imputes the data evaluated at a Mean Absolute Error (MAE) of 0.4228 m, Mean Square Error (MSE) of 0.4374 m2, Root Mean Squared Error (RMSE) of 0.6614 m, and Mean Relative Error (MRE) of 0.1063. Further, we demonstrate that the BRITS model improves the quality and reliability of sensor-generated agricultural data. Moreover, overlapping areas in the monitored field highlight the necessity of optimizing tractor paths.
A Data Collection Framework for Precision Agriculture: Addressing Data Gaps and Overlapping Areas with IoT and Artificial Intelligence
Rocco Pietrini;
2024-01-01
Abstract
The integration of sensor technology and IoT in precision agriculture has significantly improved farm operations. These technologies enable farmers to monitor crop conditions, soil moisture, and equipment performance in real-time, leading to more efficient practices. However, challenges such as missing data due to low network connectivity in farming areas and inefficiencies due to resource overuse during field operations still persist. This paper addresses these critical issues by presenting and implementing a comprehensive data collection mechanism using IoT architecture and sensors installed on farm machinery, capturing key metrics in real-time using serverless cloud technology. This data helps optimize field coverage and minimize redundant work. We further address the issue of missing data incorporating artificial intelligence-based Bidirectional Recurrent Imputation for Time Series (BRITS) model. When considering random missing data, the model successfully imputes the data evaluated at a Mean Absolute Error (MAE) of 0.4228 m, Mean Square Error (MSE) of 0.4374 m2, Root Mean Squared Error (RMSE) of 0.6614 m, and Mean Relative Error (MRE) of 0.1063. Further, we demonstrate that the BRITS model improves the quality and reliability of sensor-generated agricultural data. Moreover, overlapping areas in the monitored field highlight the necessity of optimizing tractor paths.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
