A master-sub algorithm fusion strategy for full-coverage path planning of deep-sea mining vehicle clusters

In deep-sea mining, effective path planning for vehicle clusters is crucial. Notably, few existing path-planning studies consider pipeline-related factors. This paper presents a master-sub algorithm fusion strategy for the full-coverage path planning of deep-sea mining vehicle clusters in a pipe-lifting system. The master-algorithm uses an improved DDQN algorithm to cover mining areas efficiently and avoid hose entanglement. The sub-algorithm combines a modified inner-spiral algorithm with an A * algorithm for wide-area coverage. Experiments show this strategy achieves 100% coverage in complex environments, prevents hose entanglement, and enables vehicle and hose recovery. • The spiral algorithm’s directional selection strategy is enhanced to increase flexibility and escape dead zones. • The DDQN algorithm enables four mining vehicles to cover the mining area collaboratively. Compared to DQN, it achieves full coverage with shorter paths and improves training results. • A relative angle-counting mechanism between vehicles is designed to monitor the spatial posture of transportation hoses and prevent entanglement. • A dynamic action selection mechanism is added to the DDQN to increase the probability of selecting high-quality samples. • Two reward functions are designed for deep-sea mining’s hydraulic pipeline lifting system in DDQN: one for full-coverage path strategy and another for restoring the initial positions of vehicles and hoses.