Efficient complete coverage path planning for mining vehicles in complex environments via region decomposition and finite-state control

Complete coverage path planning for deep-sea mining vehicles is difficult because large mining areas require coordinated planning across multiple spatial scales, while dense and irregular obstacles can readily induce coverage deadlocks. This paper proposes a three-tier hierarchical framework integrating workspace decomposition, inter-subregion routing, and intra-subregion coverage. First, an obstacle-feature-driven Boustrophedon method classifies obstacles into structural and local categories, allowing decomposition to be performed only with respect to obstacles that significantly affect workspace topology. Second, inter-subregion routing is formulated as an Open Traveling Salesman Problem and solved using ant colony optimization to obtain an efficient visitation sequence. Third, a finite-state-machine-driven dual-state cooperative planner coordinates coverage, transition, avoidance, and recovery through explicit state switching, supported by a multi-metric heuristic library, a right-turn-priority steering rule, and hierarchical deadlock recovery. The framework is evaluated through large-scale mining-field simulations, comparisons with Zigzag and Spiral methods, ablation and scalability analyses, and ground-based experiments using a tracked robotic platform. The proposed method achieved high empirical coverage reliability, exhibited approximately linear computational growth over the evaluated map sizes and generated trajectories with low tracking error and continuous angular-velocity variation in the ground experiment. These results demonstrate complete coverage and kinematic executability under the tested conditions.