Increased environmental pollution, costs and waiting time in the urban traffic sector are not only important issues for politics and city planners. These issues significantly affect individual private users of urban transportation, too. Overcrowded streets and transportation modes cause the desire for alternatives in transportation. The complexity of relations within urban mobility are unclear for most private users and lead to inefficient use of existing transport modes. In addition, an overview of all urban transportation options is not given.


Focus of this project is to incentivize efficient multi-modal urban transport. For this reason, a tool is developed that can assess the usage behavior of mobility options and show optimization potential. In addition, it can highlight the impact and relevance of mobility related decisions and it can eventually give specific recommendations for e.g. the purchase of vehicles or the usage of different transport modes. The project consists of two main work packages. First, a calculation model will be created that can assess the mobility behavior with regard to the target values costs, emissions and time (additional targets possible). In the end, an optimization is possible by using machine-learning methods.


Concept phase

In the first phase of the project, potential data sources and interfaces of calculation model will be identified. In addition, an extensive literature research on existing work and methods on the topic is required.

Model building

The calculation model consists of four sub-models. Each of these sub-models contains the calculation methodology for one of the target values. This results in a cost model, an emission model, a time model and further models for "soft" factors such as comfort, flexibility and others. The aim of the project is to take a holistic view of these values, so that e.g. emissions and costs are not only calculated in the direct use of transportation, but production and recycling ("cradle to grave") are also taken into account. In addition, the model should also be able to depict forecasts for the coming years.

Plausibility check of the model

The plausibility check will be carried out with the help of existing mobility data from surveys such as MiD (Mobilität in Deutschland) and MOP (Deutsches Mobilitätspanel). The calculated target values from the input data of the surveys will be checked for plausibility.

Machine learning process

An optimization methodology is to be generated using machine learning methods. With the help of the results of the preceding plausibility check, a neural network is generated that can generate and optimize its own results from new input data after sufficient learning processes.

Validation of the optimization

A validation of the optimization methodology takes place by recording and generating new mobility data. These are used as an input for the system and the results of the optimization are checked for plausibility.