Fault Detection and Fault-Tolerant Control in Multi-Agent Systems

Multi-agent systems comprise networks of autonomous or semi-autonomous units—robots, vehicles, satellites or sensors—that cooperate to achieve collective objectives. In such systems, faults in sensors, actuators or communication links can propagate rapidly, undermining stability, safety and performance. Fault detection seeks to identify the occurrence and nature of anomalies as early as possible, while fault-tolerant control aims to maintain coordinated operation in their presence through reconfiguration, redundancy or robust estimation. Modern approaches emphasise distributed architectures in which each agent uses local information and peer-to-peer communication to detect, isolate and accommodate faults without reliance on a central supervisor. Key techniques include consensus-based observers, unknown-input observers that decouple disturbances, event-triggered communication to conserve bandwidth and optimisation-based design of estimator gains. Practical applications range from unmanned aerial and ground vehicle platoons to satellite constellations and industrial sensor networks, where resilience to component failures and environmental disturbances is essential for global-scale missions.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Fault Detection and Fault-Tolerant Control in Multi-Agent Systems publication trend

The graph below shows the total number of articles in fault detection and fault-tolerant control in multi-agent systems across all publications each year (not limited to Nature Index journals).

Technical terms

Multi-Agent System: A collection of interacting autonomous units that cooperate to perform tasks or achieve shared objectives.

Fault Detection: The process of identifying the presence and type of anomaly in a system component or signal.

Fault-Tolerant Control: Control strategies designed to maintain system performance and stability in the presence of faults.

Unknown-Input Observer (UIO): An estimator that reconstructs system states and fault signals while decoupling unknown disturbances.

Linear Matrix Inequality (LMI): A convex constraint on matrix variables used to formulate and solve robust controller or observer design problems.

Event-Triggered Mechanism: A communication or computation strategy that activates only when predefined error or residual thresholds are exceeded, reducing resource usage.

References

1. Event-Triggered Collaborative Fault Diagnosis for UAV–UGV Systems. Drones (2024).
2. Distributed Fault Estimation for Time-Varying Multi-Agent Systems With Sensor Faults and Partially Decoupled Disturbances. IEEE Access (2019).
3. Neighborhood‐based distributed robust unknown input observer for fault estimation in nonlinear networked systems. IET Control Theory and Applications (2022).