Task Allocation in Agricultural Remote Sensing Applications using Submodular Maximization Algorithm

Abstract

This paper addresses the application of task allocation algorithm for agricultural remote sensing using a group of UAVs. A task allocation problem is formulated as a form of reward function maximization problem. The reward function is defined by considering the following three aspects. The first aspect is to assign tasks to UAVs with proper sensors to detect the most severe hazards for the crop raised in each field at the observation season. The minimization of the total operation time of UAVs is another aspect for short observation time and small fuel consumption. The last aspect is to assign tasks uniformly to the UAVs to prevent overloads on certain UAVs. In order to apply submodular maximization-based algorithm, the subexponential submodularity tester is applied to verify the submodularity of the defined reward function. A greedy algorithm-based task allocation method is applied to the designed task allocation problem as an example of submodular maximization-based algorithm. Simulations are conducted to show the characteristics of the designed reward function and to compare the results of the task allocation method with the optimal solutions.