An Autonomous Unmanned Bulldozer System for Coal Clearance in Hazardous Ship Hold Environments

This paper proposes an autonomous unmanned bulldozer system for hazardous coal clearance operations within ship holds. Traditional manual methods pose critical safety risks, including structural collapse, poor visibility, and rollover hazards. To address these challenges, an integrated framework is developed comprising: (1) a perception module that fuses LiDAR-SLAM with probabilistic voxel mapping to extract semantic features (bilge coal, bulkheads) and enable real-time 3D reconstruction in dynamic environments; and (2) a hierarchical control strategy featuring Dubins-enhanced navigation for efficient path planning combined with hybrid area-splitting/bull-ploughing methods for full-coverage coal aggregation. Experiments validate significant positioning accuracy improvements over LOAM algorithms achieved through multi-sensor fusion. Field tests in operational ship holds confirm the system’s capability to perform reliably under extreme conditions, while digital twin implementation substantially reduces development cycles. This work delivers a robust solution for automating life-critical coal clearance operations, offering significant enhancements to port safety and operational efficiency.