Redesign of a statistical driver model for the 3D-method

Abstract

In the field of customer use of the development of vehicles, the so called 3D method allows the representation of the customer behavior through the parameter space. Building a statistical driver to work with this method is a challenging problem given the limitations of available data and the high statistical accuracy requirements in the simulation of strongly correlated variables. Previous students have failed to group the information in a way that preserves the information available. Through this document, the different simulation strategies that were built will be described with focus on the one that proved to be the most successful. This technique is based in the creation of variable length blocks of information to be resampled en reorganized during the simulation. The statistical resemblance among the real and the simulated roads leaves us one step closer to the creation our statistical driver that will be used for simulation and optimization purposes.