Intelligent Cooperative Control for Multiple Unmanned Underwater Vehicles Using Classical Algorithms

Abstract

The use of Unmanned Underwater Vehicles (ROV/AUV) in multi-vehicle systems is a critical solution for ocean exploration missions. In this study, a three-layered deterministic “FiratROVNet” autonomy framework centered on Model Predictive Control (MPC), which offers high trajectory tracking precision, is proposed. In the first layer of the architecture, a 3D global path is generated using the A* (A Star) algorithm over known map data; in the second layer, static obstacles not present on the map are bypassed by a local planner; and in the third layer, Logarithmic Artificial Potential Fields (APF) react to suddenly appearing dynamic obstacles. Logarithmic damping solves the tumbling problem of the ROV caused by linear APF when approaching obstacles closely. Inter agent obstacle sharing is provided by Graph Attention Networks (GAT), while task handover is executed through a mathematical leader selection based on battery levels and centrality. In the developed system, the optimum force commands generated by the MPC are directly allocated to the 6 degree-of-freedom (6 DoF) thrusters of the BlueROV2 platform using Euler rotations and scalar/vectorial products. Simulation results prove that the proposed A*, MPC, and LogAPF architecture reduces trajectory error, completely eliminates collisions, and optimizes processor load compared to the classical A*, PID, and DWA architecture