Multi-source Trend Analytics for Intelligent Transportation Systems (MISTic) 

Better traffic management is essential to turn Brussels into a smart(er) city. The present project is situated in the domains of mobility, intelligent transportation systems (ITS) and smart cities. It involves Sirris (EluciDATA Lab) as the research organization and promotor, Macq as industrial partner, and VUB (ETRO and MOBI research labs) as academic partner and co-promotor.

MISTic focuses on researching the conception and validation (in the ITS domain) of an advanced trend analytics engine that facilitates the realization of dedicated solutions concerned with the acquisition of an accurate and context-aware understanding (or situational picture) of a (traffic) situation and the provision of insightful feedback to the different stakeholders involved in that situation (e.g. traffic managers, road users) and thus facilitates their decision making. In order to acquire such a situational picture, data from several sources, including sensors at fixed locations (such as cameras, license plate readers, inductive loops, etc.), sensors moving with the traffic (such as data collected by probe and connected vehicles) and other sources such as historical and weather data, need to be combined (fused) and exploited (which is known as multi-source data fusion). Then, methods enabling the profiling of the traffic situation in a context-aware way, the interpretation of such situation and the prediction of the possible evolutions of that situation need to be developed in order to provide feedback about detected relevant trends, facilitating decision making. Examples of possible trends to be considered are the number of vehicles of the same type (cars, motor bikes, trucks, trucks transporting dangerous substances, etc.) in a tunnel and how they interact, the different types of drivers in a road segment based on their driving behavior, the number of vulnerable users at crossroads, etc.

The main research objectives are (1) to investigate hierarchical and adaptive multi-source data and model integration techniques enabling the combination of information from heterogeneous sources, (2) to research knowledge propagation techniques within and across data sources, (3) to explore methods for estimating the quality of data fusion, (4) to research methods enabling context-aware situation profiling, interpretation and evolution prediction, and (5) to realize and validate a proof-of-concept trend analytics engine and demonstrate its potential via realistic use cases. Two such use cases have already been identified together with the industrial partner, Macq, during the preparation of this proposal and will be further elaborated during the project. A first use case relates to tunnel (and infrastructure) safety. Another use case revolves around the classification or profiling of road user behavior. It must be noted that the goal of this research project is to develop the technology behind a trend analytics engine. The use cases that we identified and potential new use cases that will be defined during the PhD are solely meant to demonstrate the potential of that trend analytics engine as a sort of toolkit of optimized and validated modules enabling the realization of advanced mobility applications.