A Context-Aware Model Predictive Control Approach for Multi-Site Operations of a Surface Vehicle-Manipulator System in Dynamic Marine Environments

Regular inspection, maintenance, and repair (IMR) operations are essential for marine structures. Surface vehicle-manipulator systems (SVMS), with high mobility and manipulation capability, offer a promising solution for maritime IMR tasks. However, wave-induced motions of the surface vehicle and the spatial distribution of target structures introduce significant challenges to end-effector regulation accuracy and operational stability in IMR tasks. This paper presents a context-aware model predictive control (CA-MPC) approach to address these challenges. The proposed CA-MPC framework incorporates a wave-induced velocity term into the prediction model to enable dynamic compensation for wave disturbances. Furthermore, a context-aware motion distribution strategy is developed, which dynamically adjusts the optimization objectives and actuator activation status based on operational scenarios, enabling coordinated control between the vehicle and manipulator subsystems. Simulation studies are conducted on two typical scenarios: (i) end-effector regulation under various sea states, and (ii) multi-site IMR operations. In both scenarios, the proposed approach ensures precise end-effector regulation under wave disturbances while enabling efficient transitions between spatially distributed operational sites. Results show that the CA-MPC enhances task accuracy and robustness, while the motion distribution strategy demonstrates strong adaptability, indicating its potential for large-scale offshore IMR operations.