Stochastically Modelling Road Topography Identifying Road Topography: Characteristics and Stochastically Modelling it for Simulations

Abstract

Road topography is the 2-dimensional elevation profile of a road and is a critical factor in evaluating vehicle performance and energy consumption in the transport sector. Currently the most used method is to model existing roads. This method requires that a suitable real representative road exists or needs to be constructed in expensive testing tracks. It also runs into the problem of being constrained to one road and can’t capture the larger characteristics of an area. This study aims to find alternatives to this method by modeling road topography stochastically to create an infinite number of roads with varying characteristics. To achieve this, it investigates real world road data and develops various stochastic modeling approaches to generate synthetic road profiles that capture the key topographical characteristics observed in actual roads. The models that were examined were AutoRegressive(1), ARMA(1,5) and Markov chain model. The models were designed to model the slope of the road to capture the characteristics of a set of input roads. The generated models are validated against real world data using multiple comparison metrics, to assess their ability to capture real-world characteristics. This is to evaluate the potential for simulation applications. To identify which characteristics are essential for road topography modeling, a qualitative study was performed, where experts within vehicle engineering disciplines were interviewed.