The Digital Twin (DT) concept has gained attention for its potential to enhance agricultural productivity and sustainability by creating virtual replicas of physical objects or environments. Fixed IoT installations enable the implementation of DTs as one of the main data sources; however, they introduce reliability, scalability, and cost limitations. Furthermore, these installations may miss crucial environmental changes, impacting the value of the collected data. This paper proposes an Unmanned Ground Vehicle (UGV) as an active sensing platform for enabling agricultural DTs. It integrates a mobile base with a modular sensing payload, including imaging and environmental sensors, to autonomously gather field information. The platform can replicate fixed IoT installations by incorporating a robotic arm, creating time-series datasets for comprehensive farm monitoring. Additionally, the platform can adapt to environmental changes by leveraging Edge AI and distributed learning techniques in the data collection process. To validate our proposal, we present a use case where the platform navigates through an orchard, collecting time-series data on fruits, using an RGB camera. The results from simulation and field experiments quantitatively evaluate the platform’s scalability, accuracy, and dynamicity.
Overcoming Limitations of IoT Installations: Active Sensing UGV for Agricultural Digital Twins
Vecchio, Massimo
2023-01-01
Abstract
The Digital Twin (DT) concept has gained attention for its potential to enhance agricultural productivity and sustainability by creating virtual replicas of physical objects or environments. Fixed IoT installations enable the implementation of DTs as one of the main data sources; however, they introduce reliability, scalability, and cost limitations. Furthermore, these installations may miss crucial environmental changes, impacting the value of the collected data. This paper proposes an Unmanned Ground Vehicle (UGV) as an active sensing platform for enabling agricultural DTs. It integrates a mobile base with a modular sensing payload, including imaging and environmental sensors, to autonomously gather field information. The platform can replicate fixed IoT installations by incorporating a robotic arm, creating time-series datasets for comprehensive farm monitoring. Additionally, the platform can adapt to environmental changes by leveraging Edge AI and distributed learning techniques in the data collection process. To validate our proposal, we present a use case where the platform navigates through an orchard, collecting time-series data on fruits, using an RGB camera. The results from simulation and field experiments quantitatively evaluate the platform’s scalability, accuracy, and dynamicity.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
