The optimal management of a fleet of drones is proposed in this paper for providing connectivity to sensors and actuators in Industrial Internet of Things (IIoT) scenarios. The persistent mission without any human intervention on the battery charge is obtained by means of an on-field wind generator supplying a charge station that adopts resonant wireless power transfer. The objective of the fleet management is to provide the best connectivity over the time considering the variability of both the bandwidth request and the wind energy availability. The optimal management is performed by a system controller adopting reinforcement learning (RL) for deciding the number of drones to take off and, consequently, the instantaneous provided bandwidth. A constant charge time of drone battery represents a key element of the system because this enables to strongly reduce the complexity of the system controller task. To this purpose, an adaptive current control for the charge station is introduced to compensate charge time variabilities due to the coupling factor changes caused by misalignments that can occur between a pad and a drone. The results have highlighted that the RL provides good performance improvement in case of green generation. An important aspect arose from this study is the ability of RL to increase the saved energy even if it is not considered as a target of the controller.

Green wireless power transfer system for a drone fleet managed by reinforcement learning in smart industry

Faraci G.;Raciti A.;Rizzo S. A.;Schembra G.
2020

Abstract

The optimal management of a fleet of drones is proposed in this paper for providing connectivity to sensors and actuators in Industrial Internet of Things (IIoT) scenarios. The persistent mission without any human intervention on the battery charge is obtained by means of an on-field wind generator supplying a charge station that adopts resonant wireless power transfer. The objective of the fleet management is to provide the best connectivity over the time considering the variability of both the bandwidth request and the wind energy availability. The optimal management is performed by a system controller adopting reinforcement learning (RL) for deciding the number of drones to take off and, consequently, the instantaneous provided bandwidth. A constant charge time of drone battery represents a key element of the system because this enables to strongly reduce the complexity of the system controller task. To this purpose, an adaptive current control for the charge station is introduced to compensate charge time variabilities due to the coupling factor changes caused by misalignments that can occur between a pad and a drone. The results have highlighted that the RL provides good performance improvement in case of green generation. An important aspect arose from this study is the ability of RL to increase the saved energy even if it is not considered as a target of the controller.
Artificial intelligence; Drone; Industry 4.0; Internet of Things; Wind generator; Wireless power transfer
File in questo prodotto:
File Dimensione Formato  
AE_Green Wireless Power Transfer System for a Drone Fleet Managed by Reinforcement Learning in Smart Industry.pdf

non disponibili

Tipologia: Versione Editoriale (PDF)
Dimensione 1.83 MB
Formato Adobe PDF
1.83 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.11769/374676
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 23
  • ???jsp.display-item.citation.isi??? 15
social impact