State Estimator Design for Real-time Controlled Restraint Systems

In this paper, a method is presented to estimate the motion of a vehicle occupant in frontal impact. Knowledge of the occupant´s motion during a crash can significantly reduce the risk of injury, as restraint systems could be real-time adapted to the optimal settings. It is shown that the chest acceleration of the occupant, an important measure of injury, is accurately estimated by filtering of the belt displacement at the retractor side. Low order, nonlinear models are constructed from complex models, and a state estimator filter is designed using a linearization of these low order models.