In this paper we propose a projection based observer that can be used to estimate state vectors such as position and velocity for its motion control using non-uniform interval multi-sensor outputs. The output measurement we will use in estimating the state is a series of known multi-sensor asynchronous outputs. The idea for the proposed projection based observer is that each measurement output give us a piece of information which we can use to project the state vector onto the null space of C (output matrix) in order to reduce the error of state estimation. We will explain how to construct the projection operator in a Hilbert space and show the noise rejection of this algorithm by using Lyapunov method. The applicability of the proposed based observer to SAUV (semi-Autonomous Underwater Vehicles) navigation system is demonstrated through computational analysis with theoretical data. Its comparison with traditional Kalman filtering is assessed for computer simulations of the KRISO SAUV and experiments of the FAU Ocean Voyager II Autonomous Underwater Vehicle.


There are numerous applications where the usage of an Autonomous Underwater Vehicle is desired. Born from the military requirements for stand-off weapons, remote sensing, mine field mapping and clearance, and coupled with the rapid commercialization of microcomputer technology, it is now possible to foresee some civilian use for AUVs in ocean science, offshore operations, environmental monitoring and underwater inspection. One continuing issue is the precise navigation of such vehicles underwater. While differential GPS (DGPS) is widely used for survey ships and underwater vehicle while surfaced, the navigation problem underwater requires the fusion of data from sensors such as Doppler sonar in a ground locked mode, compass data (a three axis magnetometer), acoustic systems like a SSBL and low cost inertial units.

This content is only available via PDF.
You can access this article if you purchase or spend a download.