Resilient formation control in multi-agent systems considering leader failure
Takuya Murakami, Toru Namerikawa · SICE Journal of Control Measurement and System Integration · 2025
This research investigates how to control multi-agent systems to maintain formation even when any agent, including the leader fails. We introduce cooperative control approach to solve the formation problem, ensuring that agents can maintain their relative geometric positions even in the event of a malfunction in any agent, including the leader. To achieve this, we developed an algorithm that uses the reliability of each agent, determined through fault detection using distributed observers, to limit information exchange with low-reliability agents. Additionally, we extend the adjacency matrix to allow any follower to assume the role of leader in the event of the leader's failure. This enables non-faulty agents to maintain the formation despite agent failures, including the leader. Finally, the proposed approach is validated through theoretical analysis and numerical simulations.