A Spectral Clustering Enabled Dynamic Task Allocation Approach of Multiple UAVs

In the cargo distribution field, UAVs play an important role and have wide applications. However, the distribution timeliness, flight energy consumption, and load capacity directly affect the implementation of distribution tasks by multiple UAVs in the urban scene. To cope with the above issues, this paper proposes an approach of spectral clustering enabled dynamic task allocation of multiple UAVs. First of all, we establish an objective function at the cost of mission delay penalty and UAV flight energy consumption. Meanwhile, considering the constraints of flight distance and load capacity, the optimization model of multi-UAV task allocation is generated. Moreover, based on the divide and conquer concept, we divide the task allocation into several sub-problems and use a spectral clustering (SC) algorithm for task grouping. The tasks in each cluster need to be ordered, so we can get the optimal task allocation scheme. Furthermore, due to the limitation of UAV energy consumption, the distribution task may be interrupted. Therefore, in the face of this situation, the task allocation mechanism needs to be dynamically adjusted to ensure the completion rate of tasks. The simulation experimental results in an urban environment illustrate the effectiveness of the proposed approach in dynamic task allocation for the cargo distribution issue.