An information theoretic source seeking strategy for plume tracking in 3D turbulent fields

We present information theoretic search strategies for single and multi-robot teams to find and localize the source of a biochemical or radiological materials in turbulent flows. In our work, robots rely on sporadic and intermittent sensor readings to synthesize information maximizing exploration strategies to find and localize the position of the source. By reasoning about the spatial distribution of these sensory cues, the robots are able to construct a belief distribution over the possible positions of the source. The belief distribution is then employed to synthesize motion strategies that drives the robots to regions in the workspace that results in the largest decrease in the entropy of the belief distribution for the source position. We validate the proposed strategies in 2D and 3D environments and consider the performance of the strategies when robots have limited access to global pose information. In particular, the proposed strategies are validated using a three dimensional (3D) time-varying computational fluid model of the 2010 Deep Water Horizon oil spill.