RoCCl - Road Condition Cloud

• Applicants: Professor Dr.-Ing. Roman Henze; Dr.-Ing. Mark Wielitzka, since 3/2021
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics ; Traffic and Transport Systems, Intelligent and Automated Traffic
• Term: from 2018 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 392112193

Numerous accidents in public traffic occur due to lack of knowledge about the current road condition, i.e. the friction coefficient between road and tire. Commonly, the driver can only react and adapt his way of driving highly delayed to changes in friction. The same holds true for driver assistance systems, e.g. emergency breaking function, thus on low friction surfaces only a reduction of the collision velocity can be realized instead of avoiding the accident. The project goal is to estimate the road condition using probability-based data fusion via heterogeneous sources and make it available for other traffic participants though a time and location dependent map, the Road Condition Cloud. These information can be used by the vehicles, e.g. to warn their drivers predictively or to adapt driver assistance systems.Within the vehicle a detailed model of vehicle dynamics with additional use of serial sensor data, i.e. accelerations, wheel velocities and yaw-rate, is used to estimate the present friction coefficient and a corresponding confidence value. Additionally, information of onboard and environment sensors (temperature, windshield wiper status, GPS-position, distance to other vehicles etc.) are consulted. The information is sent to the RoCCl and is preprocessed online. Therefore, large volume of data with high velocities and great variety have to be considered. Using methods of probability-bases sensor fusion heterogeneous information of the individual vehicles in one position, again considering a corresponding confidence, are considered. Additional information about temperature or rainfall are used to extend the individual estimations to a global time-variant map. Thus, the estimation confidence in one position can be increased due to information of many vehicles. Furthermore, positions, which are not stimulated directly, can be occupied with an estimation of the road condition by interpolation within the map. The communication to the cloud offers the possibility to adapt the driver assistance systems even before the trip started. Variations in time are considered as well, i.e. abrupt changes due to sudden rain or slow changes due to drying surface.The functionality of the RoCCl is presented in simulation with scalable number of vehicles. Within the simulation the information from many real vehicles are embedded to validate the results on real experimental vehicles.

• DFG Programme: Research Grants
• Co-Investigator: Professor Dr.-Ing. Ferit Küçükay
• Ehemaliger Antragsteller: Professor Dr.-Ing. Tobias Johannes Ortmaier, until 3/2021