Model Predictive Networked Control for Autonomous Underwater Vehicles Tracking Control with Sequence-Based Compensation

This paper proposes a model predictive network control framework for autonomous underwater vehicles (AUV). By leveraging the control sequence mechanism of model predictive control (MPC), it enables AUVs with a separated controller and body to effectively enhance tracking control robustness during random packet loss in networked control systems (NCS). In normal communication conditions, the method sends a multi-step control sequence with timestamps and version numbers to the mobile terminal for pre-storage. When consecutive packet losses occur due to network interruptions, the actuator can dynamically select control inputs from memory using a hybrid triggering mode that combines event-driven and time-driven approaches. This avoids the cumulative errors typical of traditional zeroorder hold or zero-input methods, which lack compensation for model characteristics. Finally, continuous trajectory tracking simulations were conducted for an underactuated AUV in a three-degree-of-freedom underwater environment under network state transition disturbances based on a Markov chain. The test results demonstrate that this control framework effectively mitigates the impact of random packet loss in the network.