Current reconstruction techniques for vehicle-pedestrian crashes involve documenting the observed vehicle damage, pedestrian injury, and crash environment. Given the complexity of the event, the typical approach for modeling these events involves subjective evaluations of the pre-impact conditions in a limited number of simulations. In order to develop a detailed understanding of how the pedestrian injuries relate to documented vehicle damage, computational techniques utilizing a robust optimization process must be developed to accurately reflect the impact event.

Since field investigations do not provide known initial conditions, the methodology was developed using a multi-body simulation package and evaluated using three laboratory experiments with cadavers. Geometry and stiffness of the vehicle were reproduced from experimental and computational investigations while the impact speed was a design variable. For the pedestrian, the anthropometry of each cadaver was provided as well as the orientation of the limbs (i.e., stance during gait phase). For pedestrian variables, the initial rotation of the body relative to the vehicle and the location of the pedestrian along the vehicle front were included.

The objective function was defined as the sum of the distance between known vehicle-to- pedestrian contact points and those observed in the model. Minimization of the objective function between experimental cadaver tests and computer simulations has been achieved through customized optimization procedures. The optimized result was considered the most likely impact scenario given the set of parameters and their associated ranges.

It was shown that, given known vehicle and pedestrian contact points resulting from an impact event, the optimized solution adequately approximated the initial pedestrian and vehicle conditions. With the methodology presented in this thesis, a new approach for the reconstruction of real-world vehicle-to-pedestrian impacts has been established.