ADAPTIVE ON LINE PLANNING ALGORITHM FOR AUVS EXPLORATION OF UNKNOWN OCEAN ENVIRONMENTS

An adaptive on-line planning algorithm for multiple underwater vehicles in coastal oceanographic missions is introduced. The algorithm is tested for a team of autonomous underwater vehicles exploring an unknown oceanic environment in which they can execute point-wise environmental measurements. The team final objective is the construction of a three-dimensional estimated map of the measured environmental characteristics of the whole marine region. Each vehicle chooses its next sampling point in order to maximally increase the confidence level of the overall estimate of the measured quantity and in an asynchronous way with respect to the other vehicles. It is assumed that all vehicles have availability of the positions and values of the all past measurements executed by the members of the team. Simulative test cases in which the objective is the estimation of the ocean temperature over a region are presented. The algorithm performance is compared with that theoretically achievable by Rapidly-exploring Random Trees search.