Automatic Uncertainty Propagation Based...

Automatic uncertainty propagation reduces the effort for computation of uncertainty and is thus a useful toolin a variety of applications. Typically, such tools utilize Taylorseries approximations (in particular linearization) or MonteCarlo Methods to perform the calculations. In this paper, wepropose the use of the Unscented Transform for automaticuncertainty propagation. A comparison between the approaches- realized in a toolbox for the MATLAB environment andillustrated in two application examples - shows that the UnscentedTransform overcomes some of the limitations of linearizationand Monte Carlo methods, providing reliable estimates of theoutput expectation and standard deviation in nonlinear problemsevaluating a reduced number of sigma points.