Camber Angle Estimation based on Physical Modelling and Artificial Intelligence

Abstract

The camber angle of a wheel represents one of the most important kinematic parameters in the study of the interaction between tire and road. It influences the forces that the vehicle exchanges with the road and consequently the vehicle's performance in terms of handling, holding and ride comfort. Control systems for these angles can improve in these aspects. Consequently, the measurement of the camber angles represents an aspect of primary importance. There are many solutions for measuring camber angles when the vehicle is not in motion, but developments must be made for measuring these angles during manoeuvres. Direct measurement of camber angles requires the insertion of sensors inside the wheel which increase complexity and production costs. This paper proposes a methodology for estimating such angles that makes combined use of a double-track vehicle model, non-linear Kalman filter and artificial intelligence. It exploits simple measures to be made on a road vehicle: longitudinal velocity, yaw rate, steering angle of the wheels and rotational speed of the wheels. The estimator was tested through three manoeuvres: a lap on the Hockenheim circuit, a slalom and a constant radius cornering. The metrics used to describe the error: are Root Mean Square Error (RMSE) and Normalized Root Mean Square Error (NRMSE).