Robust online roll dynamics identification of a vehicle using sliding mode concepts

• A robust online roll dynamics identification algorithm based on sliding modes is proposed.
• The joint state and parameter estimation requires only measurement of the roll rate.
• The mechanism is based on an adaptive gain sliding mode observer.
• A vehicle characteristics classification with respect to roll dynamics is shown.
• Evaluation of the concept is also performed in-vehicle using an experimental car.

This paper proposes a robust observer concept for joint estimation of system states and model parameters related to the roll dynamics of a vehicle. Using sliding mode concepts introduces robustness to parametric uncertainties and also allows reconstruction of the latter. These model parameters are of interest for vehicle dynamics assessment and estimation of the roll angle. A novel classification concept exploits these parameter estimates for assessing the roll dynamics. An additional benefit of the proposed method is the minimal requirement of measurement equipment as it only relies on cost-efficient angular rate and acceleration sensors. Evaluation of the framework is performed in simulations and real-world using an experimental vehicle.