Comparison of Path Planning Approaches for Harmful Algal Bloom Monitoring

Harmful algal blooms are an important phenomenon to monitor because of their potential for damaging ecosystems that depend on the affected body of water. A useful instrument to monitor these are Autonomous Underwater Vehicles with attached sensors. In this work, we explore methods to plan vehicle routes that maximize the coverage of sensor readings in algal blooms, as opposed to the traditionally used lawnmower coverage surveys. This is done in order to save time and cover blooms with large spatial extents. To this end, we focus our search on areas of high concentration, ignoring the monitored quality of low concentration areas. We also look to monitor areas of high rates of change, such as the continuous boundary of an algal bloom, to inform marine ecologists of the extent and boundary conditions of a bloom. We use Gaussian process regression to build a map of the concentration of chlorophyll and greedily plan paths using a supplied metric. Our simulation results show that a simple metric can be used to monitor the concentration of chlorophyll as well as the rate of change of the concentration. The results are validated on real and synthetic datasets.