This thesis presents the design of a pedestrian protection system, which can determine the necessity of emergency braking. During my research on the subject of driver assistance systems, I became acquainted with the environmental sensors of the vehicle industry. In my evaluation of these sensors I focused on the automotive radar. I analyzed the road traffic situations in terms of pedestrian run over. After that I investigated the test specifications of the Euro NCAP autonomous emergency braking assistant ratings. I then researched in the field of probabilistic models. When selecting from these models, I took into consideration that it has to function in an embedded system with limited computing power and exact timings. In this thesis I present an implementation of the naive Bayesian probabilistic model.
During the implementation of the pedestrian run over detection software, I used signals provided by radar and car movement monitoring sensors to determine the probability of a pedestrian run over. I also found a way to predict potentially dangerous situations, in which there may exist a pedestrian occluded by an obstacle. Using a data set preprocessed by myself the system parameters were adjusted and the performance was evaluated. Finally, I present the opportunities for further development.