A Two-Phase Task Allocation Strategy With a Hybrid Architecture

• Published in: 2024 27th International Conference on Computer Supported Cooperative Work in Design (CSCWD) pp. 55 - 60
• Main Authors: Zhang, Xiaoyu, Wang, Wen, Ren, Shuangyin, Gong, Xiaomin, Yang, Yuxuan, Wang, Jingchao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 08.05.2024
• Subjects: Autonomous aerial vehicles; Clustering algorithms; communication overhead; Computer architecture; Federated learning; heterogeneous UAVs; hybrid architecture; Real-time systems; Robustness; Scalability; task reallocation; Two-Phase task allocation
• ISSN: 2768-1904
• DOI: 10.1109/CSCWD61410.2024.10580586

In complex disaster relief or unmanned delivery scenarios, the collaboration of heterogeneous UAVs faces numerous difficulties with dynamic and harsh environments, such as limited resource capacity and communication constraints. Especially for task allocation, the computational and communication burden poses many challenges for real-time task assignment and execution. To achieve a high allocation efficiency, this paper presents a Two-Phase task allocation strategy within a hybrid architecture. In the centralized phase, considering UAVs' task requirements and capability, an initial task allocation is made from an overall perspective. In this phase, it determines the suitable types and numbers of UAVs, which guarantees the completion of tasks and reduces the potential communication to the greatest extent. In the distributed phase, specific UAVs within type-specific clusters are selected for executing tasks considering individual capability and physical location. This reduces the risk of single-UAV failures, enhancing the robustness and scalability of task allocation. Additionally, during the distributed phase, an incremental update method is employed to reduce communication latency and resource consumption. Experimental analysis and comparisons with the existing approaches demonstrate that our approach effectively reduces communication overhead and significantly improves task allocation efficiency. Furthermore, in the event of UAV failure or malfunction, it allows for a swift reallocation of tasks to other available UAVs.