Safety-Constrained Model Predictive Trajectory Tracking Control for Self-Reconfigurable AUVs with Energy Optimization

Self-reconfigurable autonomous underwater vehicles (SRAUVs) exhibit significant potential for complex underwater missions due to their adaptability and reconfigurability. However, existing trajectory tracking controllers often fail to address the coupled challenges of model uncertainties, structural safety constraints, and energy efficiency in dynamic reconfiguration scenarios. To bridge this gap, this paper proposes a safety-constrained model predictive control (MPC) framework integrated with radial basis function neural network (RBFNN) compensation and quadratic programming (QP)-based energy-optimal control allocation. First, a comprehensive kinematic-dynamic model incorporating thruster dynamics and link force constraints is established. Subsequently, a RBFNN-enhanced model predictive controller is designed to compensate for system nonlinearities and external disturbances, while a QP-based allocation strategy optimally distributes thrust forces under actuator saturation and structural safety constraints. Extensive simulations demonstrate that the proposed framework reduces trajectory tracking error and energy consumption compared to baseline MPC, while ensuring all link forces remain within safe thresholds. This work provides a systematic solution for enhancing the operational reliability and endurance of SRAUVs in real-world applications.