The CRASH project aims the research in the area of Data Science, 3D Simulation, Artificial Intelligence and Software Engineering, proposing the development of a fully functional prototype that provides a high-precision virtual reconstruction, description and diagnosis of the causes of road accidents. The project envisages the development of a 3D scanning system for vehicles, with a precise definition of the deformations suffered during an accident, and the development of an accident scene simulation system, with the identification of all forensic evidence of accidents and their measurement definition. The implementation of artificial intelligence models will provide the system with the ability to relate 3D objects and evidence from other sources of information (sensors), allowing the reconstruction of the entire timeline of the accident, combining data and relevant 3D objects, and making it possible to include at any time in the timeline mechanical and dynamic models of relevant vehicles, related to any type of 3D object. Accident reconstruction will be based on the definition of mechanical equations of kinematic and dynamic behavior of physical models resulting from 3D digitization. It is also foreseen the integration of the information registered in the vehicles to feed the dynamic behavior of the vehicle and the behavior of the driver.