Using vehicle fleet data to quantify the risk of critical driver surprise in traffic

Purpose and goal:

The project aims to develop big data analysis methods suitable for precise identification of when and where drivers actually need help from the vehicle and when they do not, to avoid or at least miminize the risk for ¨cry wolf¨ problems in future advance driver assistance systems. Expected results and effects:

Based on the concepts of comfort and surprise, we expect to find one or more viable methods to quantify these concepts in conflict situations in large-scale vehicle data, and to do this in such a way that it becomes possible to decide whether the car should intervene or not based on both real, and (from the driver´s perspective) perceived, risk.

Approach and implementation:

A selection of big data analysis methods will be applied and evaluated based on data from a variety of traffic conflicts, identified in Volvo´s own driving data.