Public safety is one of the most pressing concerns in the cities of today. Whenever multiple citizens gather in public space, crowd dynamics can cause hazardous situations. To avoid evacuations and disasters, it is pertinent that the movement and evolution of a crowd can be tracked, predicted and adapted, while assuring privacy. Therefore, this project will integrate crowd tracking and modeling, which will give input to a routing algorithm. By dynamically calculating the optimal routes, based on the predicted crowd state, a flexible system is built which can anticipate an unsafe evolution. The resulting proof of concept can be embedded in a smart city infrastructure to keep its citizens safe.