Estimation of space-speed profiles : A functional approach using smoothing splines

Résumé

In this talk, we will present a functional modeling approach of space-speed profiles (speed as function of vehicle position). Due to uncertainties in the measurements of position and speed from sensors, the problem is equivalent to a nonparametric regression model. A smoothing spline model using derivative information based on the theory of reproducing kernel Hilbert space (RKHS) will be presented. Then, after a curve registration step, the construction of a functional boxplot based on the center outward ordering induced by band depth will be shown on a set of space-speed profiles.