Where do you sample? - An autonomous underwater vehicle story

We present an experimental evaluation of various sampling path strategies for an Autonomous Underwater Vehicle. Both systematic and stratified random sampling path strategies were evaluated based upon their estimation accuracy for isotropic and anisotropic scalar fields, as well as the relative energy consumption. We present results from several experimental trials that shows that the stratified random sampling strategy minimizes estimation error for denser sample distributions, and the systematic sampling strategy minimizes estimation error for sparser sample distributions. Finally, we experimentally show that the systematic spiral path sampling strategy is the most energy efficient.