Robust Multi-Model Tire-Ground Force Estimation Scheme

Abstract

A new approach to estimate the tire-ground forces at each tire position is provided in this paper. Our method exploits two model realizations, on the one hand, the famous bicycle model is used to estimate the longitudinal/lateral tire-ground forces at each axle. On the other hand, the hoverboard model is introduced to compute analogous forces at the left and right virtual tires. The algorithm is capable to correctly split the tire-ground forces given by both models without the necessity of a tire representation, the longitudinal slip, and the side-slip angles. The precise tire-ground forces computation at each tire position is made with a closed-loop observer scheme using a constrained extended Kalman filter. The results are validated with three tests: two slaloms, the first one performing within the linear tire operation domain and during the second test the tires are acting under the non-linear saturated region, and the third test is a circle maneuver with constant steering and longitudinal acceleration.