Information-Driven Quantum Path Planning for Unmanned Surface Vehicles in Stochastic Marine Environments

Autonomous water quality monitoring using unmanned surface vehicles (USVs) requires effective path planning under time-varying hydrodynamic conditions and spatially evolving pollutant distributions. Existing approaches often rely on static environmental assumptions or geometric shortest-path optimization, limiting their effectiveness in dynamic sensing tasks, where both information gain and environmental uncertainty must be considered. This paper proposes a unified information-driven quantum path planning framework for adaptive marine monitoring. A coupled stochastic differential model is first introduced to represent ocean currents and wind-induced drift with multi-scale temporal dynamics. Based on this model, a dynamic information field is constructed by integrating pollutant concentration, spatial gradients, and epistemic uncertainty to guide active sampling. The monitoring task is formulated as a dynamic orienteering problem with energy and redundancy constraints. To enable efficient optimization, the continuous planning problem is discretized into a QUBO formulation and solved using a QAOA-based hybrid quantum–classical algorithm with receding-horizon replanning. This design allows the planner to adapt online to evolving environmental states while maintaining computational tractability. Simulation results across multiple marine scenarios demonstrate that the proposed method provides a balanced trade-off among information gain, constraint feasibility, redundancy, and computational cost, achieving near-greedy sensing return with substantially fewer mission violations than greedy informative search, and outperforming classical metaheuristics on information acquisition under the same simulated replanning budget.