Image Processing and Model-Based Spill Coverage Path Planning for Unmanned Surface Vehicles

Remote sensing technology and unmanned surface vehicles (USVs) have great potential for spill coverage applications. However, there is still a lack of research on adequate path planners for USVs based on remote sensing images. In this study, we propose an image processing and model-based path planner that can generate an efficient path for a USV to cover a spill according to aerial images from remote sensing. We mathematically formulate a spill processing model of the conceptual USV, which can remove the spill by suction mechanism, to determine its speed limit and operation range. To perform coverage with high completeness to the spill in the workspace, we first develop an image segmentation strategy and partition the area with geometric tessellation. We then formulate the coverage problem with tessellation as a traveling salesman problem (TSP) and utilize the self-organizing map (SOM) approach for effective path planning. Finally, with real aerial images containing spills, we demonstrate the effectiveness of the proposed path planning method.