MobiSense

The project

MobiSense aims at efficiently creating spatio-temporal information on road surfaces and their wear, based on opportunistic sensing on board vehicles that are on the road for other purposes. The project is funded under grant HBC.2017.0155 by imec and Vlaio (ICON scheme). Ghent University, the University of Antwerp, and KU Leuven collaborate as scientific partners, wheras ASAsense, Be-Mobile and Qweriu develop the application-oriented part.

The project solves several important challenges:

  • Data collection

    Collecting data from large numbers of vehicles and efficient big data processing;

  • Data analytics

    Removing confounders and modifiers such as vehicle speed, differences in vehicle transfer functions, music and speech inside the vehicle, sound from other vehicles etc.;

  • Information extraction

    Converting measured quantities to applicable knowledge for road administrators and environmental agencies;

  • Action planning

    Embedding this knowledge in environmental assessment and road maintenance assessment.

Technology

Noise and vibration data is collected using proprietary hardware and software from a large number of cars that or on the road anyhow: we work with stakeholder fleets, car sharing, etc. On the server, trips are mapped to road segments and lanes and the abundancy of data is used to derive useful information about road segments such as rolling noise category, macrostructure, and waviness. Dedicated expert systems further process the information to knowledge on indicators of wear, pothole proficiency, or basic information on road pavement and speed bumps.