Road-Tire Friction Estimation for AFS Vehicle Control

Abstract: Real time information about the current road surface (friction coefficient) would be of great help for most of the vehicle systems of today. Vehicle stability systems like ABS and ESP would perform considerable better with friction knowledge. Dynamic All Wheel Drive torque distributors would, with friction knowledge, substantially enhance the behavior of the vehicle. In this thesis an enhancement of vehicle dynamic control with Active Front Steering was looked upon. The vehicle dynamic control stabilizes the vehicle with electronic steering interventions at the vehicle's front axle. The thesis goal was to enhance the stabilization algorithm by using knowledge of the road-tire friction coefficient as a system input. Two types of friction estimation algorithms were investigated, implemented and tested in a prototype vehicle. The most promising result gave the approach called slip-slope based friction estimation. The hypothesis is that the friction curve's initial slope contains information about the road-tire friction coefficient. With a parallel running gravel/rough road detector, a distinction between three different friction levels is definitely possible. The second approach was friction estimation with the brush model. The brush model is a physical model for describing the tire forces generated during driving (or braking). The road-tire friction coefficient is explicitly included in the model and that makes it very suitable for different types of parameter estimation algorithms. The estimation result, when feeding the algorithm with "good" (perfectly matching estimation data with low noise levels) data, was very satisfying. But usually are the algorithm input data, during normal driving, not very good. Though, friction estimation using the brush model is an interesting approach for the future. Better sensor signals (less noise) and more work with the algorithm details could probably result in a good and reliable friction estimator.