SINUS: Sensor Integration for Urban Risk Prediction

The SINUS project researched the options available for linking mobile human sensor technology with heterogeneous data interfaces of urban data ecosystems in a way that would enable improved forecasts for traffic risks for cyclists and pedestrians in urban road networks. Combining previously isolated data sources in a forecasting model should make it possible for forecasts on traffic risks occurring in trainable, differentiated standard traffic situations with a high degree of geographical and temporal resolution. These types of risk forecasts are intended to help make urban road traffic safer for cyclists and pedestrians.

In the city of Salzburg, which served as a test bed for the project, stress moments of cyclists were recorded over several months, with these data then used together with accident data, weather data, infrastructure data and a simulation model for cyclists and pedestrians as input parameters for training a risk forecast model. A web service was developed based on this model that provides a risk assessment for a queried time and street edge of a digital road network graph. The model was trained using data from the city of Salzburg and can also be used for cities with similar infrastructures.

After development, the model was tested within the scope of two demonstrators. These were a smart city dashboard, which can be used to view the risk assessments of a road network and in which various planning scenarios and their impact on the forecasted risk can be tested, and a smartphone-based warning application for cyclists, which transmits corresponding warning signals to users when approaching a road section that has a high risk assessment. Different options (haptic, visual, auditory) for signalling cyclists were compared and evaluated in a study involving 14 participants.

This revealed that acoustic warnings achieved the best perceptibility and provided the least distraction for cyclists.

The applicability of the forecasting model developed and the risk warnings derived from this and acceptance of these, as well as their impact on road behaviour, were examined in a final evaluation study involving 40 test subjects. The indications of risk provided generally had very little impact on the test subjects' road behaviour. However, 57% of respondents stated that the warnings increased their situational awareness, while a majority found the warnings helpful (62%) and understandable (67%). There were no negative effects identified with the warning messages during the journey, such as additional uncertainty or increased stress among cyclists.

Consortium leader

• Trafficon – Traffic Consultants GmbH

Additional consortium partners

• Kompetenzzentrum – Das Virtuelle Fahrzeug Forschungsgesellschaft mbH
• Know-Center GmbH; Universität Salzburg, Interfakultärer Fachbereich für Geoinformatik – Z_GIS
• Spatial Services GmbH