Stochastic localization of sources using autonomous underwater vehicles

We propose a new method for locating the source(s) of a fluid that is diffusing into sea water. Our method utilizes multiple Autonomous Underwater Vehicles (AUVs) whose motion is controlled via a discrete-time Markov Chain Monte Carlo (MCMC) algorithm. The MCMC algorithm relies only on local measurements of the concentration of the fluid to construct and estimate of the concentration field over the search domain, and hence localize the source(s). We prove the existence of an invariant measure for the Markov chain that is generated by the closed-loop motion of the vehicles. The convergence rate of the Markov chain is investigated through extensive numerical simulations.