A fair and effective driver rating system for developing regions

Rash driving is one of the primary reasons for road fatalities. However, the existing literature seldom talks about precise quantitative measures to assess behaviors like rash driving during traffic analytics. Moreover, the subjective factors like broken road patches, high peak bumps (or some time known as speed breakers in some countries), poor night illumination etc. have not been taken into consideration in the existing driver rating systems. This paper develops a methodology for fair and effective driver rating using a dynamic time warp (DTW) method, based on their behavioral observations as captured by smartphone sensors (GPS, accelerometer, light sensor etc.). The value of the correlation coefficient between the system rating and a manual rating is found to be more than 70% for our proposed system, SafeDrive, when tested extensively for bumps along a 7km city road patch. The system can be extended to provide a comprehensive driving score by including other road features like pot holes, sharp turns etc., in addition to bumps.