Intelligent AgUAV Path Optimisation Using SAC for IoT‐Enabled Agricultural Monitoring and Data Collection

ABSTRACT

In the context of smart agricultural monitoring, the issue of complete coverage of the field area and connectivity among agricultural Internet of Things (AgIoT) devices for data collection represents a major challenge. To resolve this issue, it is imperative to implement sophisticated optimisation methods. This paper proposes a hybrid scheme combining a multiplicatively weighted Voronoi diagram (MWVD) and deep reinforcement learning (DRL) for agricultural field coverage‐based agricultural unmanned aerial vehicle (AgUAV) data collection. The first method divides the field to be monitored into several areas of interest based on epidemics, areas with low irrigation or large infestations. The second method, which is a DRL algorithm, optimises the position of drones as part of their data collection mission at preselected agricultural cluster head (AgCH) nodes in the MWVD agricultural field. The objective of this article is to optimise the movement of drones to collect data from AgIoT sensors in the agricultural field by using the MWVD algorithm to segment the field and initialise the position of the rendezvous point AgCH collector and DRL to optimise the position of drones to reduce the total energy consumption of drones, collect data reliably and optimise the total coverage of the field in an intelligent manner.