Accurate travel time prediction models play a key role for individual driver assistance systems and advanced traveler information systems (ATIS) in general.

The two main effects expected from reliable travel time predictions are the support for the road user to make better decisions regarding route choice and improving utilization of the road network as a whole.

This project presents the design and implementation of a travel time predictor not just for average or normal traffic situations. The complete research focuses on non-recurrent traffic situations in road networks, which do not occur regularly or where road users are unfamiliar with. Three example situations will be used for the research: accidents, road works and large (social) events. The travel time predictor should be able to handle these situations with a certain degree of reliability.

For the prediction method an extrapolation type of method, the artificial neural network approach is chosen. The design of the ANN prediction model(s) is going to be “case and place” sensitive, important in the case of urban network dynamics, particularly during non-recurrent situations.

The testing platform of the travel time predictor is the microscopic traffic simulator Paramics, using its API for necessary network manipulation and data acquisition. Using a simulation environment enables the reasonable and problem-oriented analysis of the research problem.