Fault-tolerant model predictive control for unmanned surface vehicles

Unmanned surface vehicles (USVs) require robust control systems capable of adeptly compensating for potential faults to ensure operational safety and successful task execution. Addressing this requirement, we present a novel approach for computing control inputs of USVs under fault-prone conditions. Our method leverages a mathematical model, specifically a linear stochastic discrete-time model that characterizes the USV subject to actuator faults. Central to our approach is the integration of an adaptive Kalman filter (AKF) with a forgetting factor into model predictive control (MPC). This fusion enables our proposed method to effectively manage actuator faults on the USVs. The essence of our fault-tolerant control strategy lies in utilizing the AKF within the MPC framework to predict both the stochastic system model and the actuator fault parameters. Through rigorous evaluation, we demonstrate the effectiveness of our proposed method in managing actuator faults on USVs. The results highlight its capacity to ensure operational continuity and task completion even in the presence of faults, demonstrating its significance for enhancing the resilience of USV control systems in real-world scenarios.