Dual-Loop Safety Control for Underwater Vehicles: Obstacle Avoidance and Disturbance Rejection

Underwater vehicle systems face significant safety challenges from complex terrains, obstacles, and external disturbances such as ocean currents. This article presents a novel dual-loop safety control framework that addresses these challenges while ensuring trajectory tracking under actuator constraints. The framework features an incremental model predictive control outer loop for optimal trajectory generation and a finite-time convergent integral sliding-mode inner loop for robust disturbance rejection. A continuous adaptive sliding-mode disturbance observer provides real-time estimation of external forces, enabling proactive compensation. To guarantee safety under thruster saturation, exponential control barrier functions are integrated through quadratic programming, minimally modifying the nominal control while enforcing collision avoidance constraints. The proposed approach uniquely combines predictive control, robust estimation, and safety certification in a computationally efficient architecture. Experimental validation on a modified BlueROV2 heavy platform confirms that the framework maintains safety constraints while achieving precise trajectory tracking, demonstrating its effectiveness for autonomous underwater operations in uncertain environments.