Combining random and data-driven coverage planning for underwater mine detection

Abstract

Unmanned autonomous vehicles are proving themselves to be effective means for conducting underwater mine hunting missions. The resulting efficiency, reduced search time, and covert search possibilities will facilitate larger mission areas requiring many agents searching for significant lengths of time (e.g., many weeks or months). In search areas of this scale, complete area coverage may not always be feasible. Therefore, this research investigates a path planning scheme for incomplete coverage. This scheme divides the search area into cells and surveys each cell using a conventional line-sweep pattern with a row spacing that is larger than the sensor footprint. The rows of the line-sweep pattern are randomly spaced near the boundaries of each cell to decrease the probability of missing a line of evenly spaced mines. The spacing of the rows near the center of each cell are specifically determined from estimated possible mine locations. Bounds placed on the row spacing limits the amount of uncovered area to maintain an acceptable probability of detection. This method results in a probability of missing a mine that is less than the percent of unsearched area.