Void Area Detection for Efficient UAV Sensing in Radiation Dose Mapping

This paper reviews the validity of genetic algorithms for UAV path planning to map radiation doses near residential areas. In this problem domain, the measurement points during a UAV flight are not always uniformly distributed, and thus most of the mapping result require interpolation, partly for local regions where the density of measurement points is not adequate. A void area in this study is characterized as the space without sufficient measurement points within a search radius, based on the observation results for the previous UAV flight. We compared two algorithms for void area detection based on (a) the number of measurement points within the search radius and (b) the proposed method using the standard deviation of the measured values within the search radius compared with those from the UAV flight. The experimental results for a difficult-to-return zone caused by a nuclear disaster indicated that the standard deviation for void area detection was effective at identifying void areas for seamlessly finding additional measurement points for second flights.