Inverse Estimation of Friction Coefficients of Winter Road Surfaces

Abstract

This study is an extension of a method developed in a previous paper. The major concern remains one of how to estimate the friction coefficient of a winter road surface indirectly with the use of vehicular motion data. Similarly, in this paper, there is no change in the central structure of the argument; the friction coefficient is estimated as the solution to an optimization problem in which a tire model describing the interaction between tire and road surface is integrated into a genetic algorithm. The tire model differs from the previous method. The one-degree-of-freedom (1-DOF) model that formulated only longitudinal motion is replaced by a 3-DOF model, in which lateral and angular motions have also been taken into account. This revised method was applied to data measured at three sites: at intersections on a test track, on curved sections of the same test track, and at intersections on arterials in Sapporo, Japan. The friction coefficients estimated by the method were in relatively good agreement with those actually measured. Lateral and angular motions have contributed to the improvement.