Visual Analysis of Car Fleet Trajectories to Find Representative Routes for Automotive Research

Abstract

Testing is an important and wide spread practice in the development of automotive components. For the design of test methods two types of input data are often considered: (1) load data gathered from real life vehicle fleets, and (2) information of the driving routes based on road features. The development of new technologies is though complicated not only by the need to join those two data sources, but also by the too limited knowledge of the parameters and their useful combinations. As a result, information about representative driving profiles is needed. To address these problems we present a visual analytics approach for analyzing multivariate trajectories as a combination of vehicle's location and road elevation data. Our system combines trajectory clustering, intervalbased user-driven trip segmentation, and frequent sequences analysis, supported by contingency table and interval-based Parallel Coordinates visualization and enables the expert user to find representative driving profiles for the definition of very compact test courses.