(An) aggressive-driving detection method using inertial measurement unit

Abstract

Because of traffic accidents, our society suffers a lot of human and material loss. According to the statistics materials, aggressive-driving behaviors are major cause of traffic accidents. So as we can correct aggressive-driving behavior, the traffic accidents can be reduced. In this paper, we proposed the aggressive-driving detection method using Inertial Measurement Unit (IMU) and established the detection system using embedded board. We supposed to detect five aggressive-driving behaviors such as abrupt acceleration and deceleration, aggressive right, left turn and aggressive U-turn. The algorithm uses the thresholding method to detect abrupt acceleration and deceleration and utilizes Dynamic Time Warping (DTW) and k-Nearest Neighborhood (k-NN) to detect aggressive turning events. To get data pattern, we performed the car experiments. And using these data, we can make a training sets. Using twenty of training sets and fifty-four of test sets, the experimental results are 96.29% accuracy and other contributions is verified.