Distributed fault-tolerant strategy for networked robots with both cooperative and reactive controls

The paper presents a controller-observer architecture with decentralized fault tolerance strategy for a networked team of autonomous robots. In the proposed architecture, each robot runs a decentralized observer to estimate the overall system state and a distributed control law for centroid and formation tracking control. The decentralized observer is used to compute a set of residuals in charge of detecting and isolating actuator faults affecting the other robots, even if not directly connected. The fault detection strategy is designed so as to distinguish between actuator faults and activation of local reactive controls e.g., to manage the presence of obstacles. The strategy is validated via numerical simulations.