Achieving Robust and Efficient Consensus for Large-Scale Drone Swarm
Wu Chen, Jiajia Liu, Hongzhi Guo
Abstract
Achieving consensus is a crucial problem for large scale drone swam to perform collaboration tasks. There are some critical issues for the consensus of large-scale drone swarms, such as limited-time convergence and high robustness against drone failure due to stringent requirements of mission cycle and hostile environment. Note that traditional consensus cannot guarantee limited-time convergence and a robust consensus against node failure. Existing finite-time consensus faces challenges of strict prerequisites and high complexity. Moreover, leaderless and leader-follower consensus adopts different models respectively. Toward these ends, a unified consensus model for both leaderless and leader-follower modes is proposed. It achieves limited-time convergence by using distributed energy minimization. A dynamic spanning tree algorithm is designed to ensure consensus under dynamic topology. Furthermore, a robust method against node failure is proposed by combining grey prediction, average consensus, and Molly-Reed criterion. Simulation results show that the proposed methods can be adopted in both leaderless and leader-follower situations with advantages of limited-time convergence and high robustness.