An advanced methodology for truck rollover prediction based on driving situation model

This paper proposes a new method for truck rollover prediction based on simple driving situation models. The traditional methods for rollover prediction generally use complete truck models. Nevertheless, rollover situations are not tripped in some cases because of uncertainties on the model parameters, which entail serious safety threat for trucks drivers. The multi-model approach (every driving situation is represented by a simplified model) can be an alternative to reduce the complexity of truck models and at the same time to focus on the request dynamics. Whatever the situation considered, the unknown dynamic state is reconstructed by using sliding mode observation technique. The computed information leads to detect rollover risky situation on the basis of a criterion. The performance of the method developed is evaluated by simulation. The simulation results are compared to a commercial simulator of truck dynamic (PROSPER).