Erosotasunean oinarritutako gidari estiloaren identifikazioa adimen artifizialaren bitartez

Abstract

Research on improving the comfort of passengers while driving is becoming more popular. This study introduces a first step into driving style identification from comfort perspective. First, a simple threshold method is tested in a variable that quantifies motion sickness to separate uncomfortable and comfortable driving. Secondly, a more complex tree-based algorithm is trained and tested with the same purpouse. The obtained results are favourable: although the threshold method obtains good results, the tree-based algorithm shows a significant improvement, and is very effective in recognizing comfortable/uncomfortable classes with little data. These promising results pave the way for a more complex experimentation that follows the same work.