Large-Scale Multi-UAV Task Allocation via a Centrality-Driven Load-Aware Adaptive Consensus Bundle Algorithm for...

Area of Science:

• Robotics and Autonomous Systems
• Distributed Artificial Intelligence
• Swarm Intelligence

Background:

• Large multi-UAV systems face challenges in distributed task assignment due to time-varying communication and heterogeneous platforms.
• Classical methods often lead to excessive message passing and suboptimal task-resource matching in complex swarm operations.

Purpose of the Study:

• To propose CLAC-CBBA, an enhanced Consensus-Based Bundle Algorithm (CBBA) for scalable and robust task allocation in large heterogeneous UAV swarms.
• To improve upon existing distributed task assignment methods by integrating swarm-inspired self-organization and load-aware self-regulation.

Main Methods:

• CLAC-CBBA employs a centrality-based adaptive cluster-reconfiguration (CenCluster) to partition networks and reduce communication.
• A load-aware cluster self-regulation (LCSR) mechanism uses K-medoids clustering and real-time load balancing for dynamic cluster adjustments.
• CBBA bidding is localized within clusters, with synchronization of winners/bids by anchors and representatives for global consistency.

Main Results:

• Simulations demonstrate CLAC-CBBA significantly reduces communication overhead and runtime compared to CBBA and other variants.
• The algorithm achieves statistically significant improvements in total task score across various network densities and swarm sizes.
• CLAC-CBBA exhibits enhanced scalability and robustness in large-scale heterogeneous UAV task allocation scenarios.

Conclusions:

• CLAC-CBBA offers a biomimetic, decentralized approach to task assignment in large UAV swarms.
• The proposed method effectively balances scalability, robustness, and assignment optimality for complex multi-UAV missions.
• CLAC-CBBA represents a significant advancement in distributed task allocation for heterogeneous robotic swarms.