Driver Evaluation and Identification Based on Driving Behavior Data

Abstract

Driving behaviors include acceleration, deceleration, steering and lane changing. People usually use these driving behaviors to analyze drivers. Each driver has his unique driving behavior data due to their different driving habits. So we can evaluate and identify drivers by analyzing their own driving behavior. In this article, we extract ten dimensional feature of each kind of driving behavior in time domain and frequency domain, and then analyze these features by statistical analysis. Then, these features are used to establish 140 feature-statistical-distributions maps for each driver. These maps depict the driver's driving habits from all dimensions, so we call it driver's driving fingerprint. Then we calculate the difference between each driver and the driver group, so as to quantitatively evaluate the performance of the driver's behavior. At the same time, we use a part of driving behavior data to establish a fingerprint database of driving behavior for all drivers. Then we use another driving behavior data to collect the driving behavior fingerprint of each driver and match it with the database. We identified the driver with the highest matching degree as the result of recognition. The experimental results show that the recognition rate reaches 100% in our database.