Predicting Hazardous Driving Events Using Multi-Modal Deep Learning Based on Video Motion Profile and Kinematics Data

Abstract

As the raising of traffic accidents caused by commercial vehicle drivers, more regulations have been issued for improving their safety status. Driving record instruments are required to be installed on such vehicles in China. The obtained naturalistic driving data offer insight into the causal factors of hazardous events with the requirements to identify where hazardous events happen within large volumes of data. In this study, we develop a model based on a low-definition driving record instrument and the vehicle kinematic data for postaccident analysis by multi-modal deep learning method. With a higher camera position on commercial vehicles than cars that can observe further distance, motion profiles are extracted from driving video to capture the trajectory features of front vehicles at different depths. Then random forest is used to select significant kinematic variables which can reflect the potential crash. Finally, a multi-modal deep convolutional neural network (DCNN) combined both video and kinematic data is developed to identify potential collision risk in each 12-second vehicle trip. The analysis results indicate that the proposed multi-modal deep learning model can identify hazardous events within a large volumes of data at an AUC of 0.81, which outperforms the stateof-the-art random forest model and kinematic threshold method.

Safety-Critical Event Identification on Mountain Roads for Traffic Safety and Environmental Protection Using Support Vector Machine with Information Entropy

“…Advanced models, such as BPNN and CNN, generally have higher requirements for application conditions. For example, a large number of samples are required for training to improve generalization [27,28]. Basic models, such as SVM, are generally not so harsh on the application conditions.…”

“…A CNN is a type of feed-forward neural network that includes convolution calculations and has a deep structure. It is often used in feature extraction and identification of events in traffic [28,29]. The algorithm learns effectively from samples of the corresponding features, avoiding the complicated feature extraction process.…”

“…A CNN is a type of feed-forward neural network that includes convolution calculations and has a deep structure [28]. The input data of CNN is processed by convolution and pooling of multiple convolutional layers.…”

Driver Profiling Using Long Short Term Memory (LSTM) and Convolutional Neural Network (CNN) Methods

“…Examining the studies using CNN in driver profiling [27]- [28]- [29] to classify the driver behavior based on driver images taken in the vehicle, the unsafe behavior during driving condition could be detected. Gao et al [30] studied to detect dangerous driving situation using video information captured from the vehicle camera. Wang et al [31] proposed two methods, one using smartphone accelerometer and gyroscope sensor, they forecasted vehicle speed with an LSTM network, while traffic light, stop lines and crosswalks were detected using CNN method.…”

Visual Counting of Traffic Flow from a Car via Vehicle Detection and Motion Analysis

“…For the motion information processing of driving video, Kilicarslan et. al has proposed the motion profile as a spatio-temporal map to examine the vehicle trajectories [9] for vehicle identification [10], collision alarming [11,12], and pedestrian detection [13,14]. A motion trace longer than optical flow provides the stable motion of targets and can be used for object recognition because of the mechanical motion of vehicles, pedestrians, and static background.…”