Preliminary results on the active pose estimation of underwater vehicles from range measurements

Recent studies relative to range-only localization in underwater robotics applications have focused primarily on (nonlinear) observability analysis. In particular, local weak observability is guaranteed as long as the motion of the system to be localized is sufficiently rich, i.e. persistency of excitation conditions need to be satisfied. These conditions typically depend on the initial state of the system and on its inputs. Once that the challenging problem of identifying the necessary initial state and input constraints for guaranteeing local weak observability has been solved, state estimation can be performed resorting to linear (Kalman-like) filters. Yet the convergence of such state estimation approaches is local. Building on recent results in this area, this paper addresses the problem of designing input signals for a class of underactuated underwater vehicles allowing global range-only pose estimation.