A Novel Driving Behavior Learning and Visualization Method With Natural Gaze Prediction

Guoliang Yuan,Jinjia Peng,Yafei Wang,Xianping Fu

doi:10.1109/access.2021.3054951

Guoliang Yuan, Jinjia Peng + Show 2 more

Open Access

https://doi.org/10.1109/access.2021.3054951

Copy DOI

Abstract

Driving behavior analysis is vital for the advanced driving assistance system, aiming to improve driving behavior and decrease traffic accidents. Most existing driving behavior learning methods focus on either vehicle sensor information or driver's attention information, and provide a classification result on the current time data samples. The visualization of driving behavior on time series data samples could give an understanding and review of the driver's continuous actions. However, there has been little progress in combining the multi-modal vehicle and driver information on driving behavior learning and visualization. A multi-information driving behavior learning and visualization method with natural gaze prediction is proposed in this paper, which automatically integrates driver's gaze direction estimated from face camera, and various vehicle sensor data collected from on-board diagnostics (OBD) system. To accurately estimate the eye gaze under large head movement, a novel head pose-free eye gaze prediction method without calibration is proposed based on global and local scale sparse encoding, which treats the direction mapping as small gaze region classification. To understand driving behavior more intuitively, the latent features that represent different driving behaviors are extracted by FastICA from the fused time series data, and mapped into RGB color space for distinguished visualization. Experimental results demonstrate the effectiveness of the proposed method, and show that the proposed method performs better than the compared methods.

Highlights

According to the technical survey, unsafe driving behavior is the main factor in today’s traffic accidents [1], [2]
Driving data, such as steering angle, brake position rate, acceleration, and velocity, were collected for the feature clustering and proactive driving behaviors reproducing. They used nine colors to represent the clustering results on acceleration and deceleration. Unlike these previous methods involving the visualization of driving behavior on steering wheel and velocity, this paper focuses on visualizing multi-modal fusion data from both vehicle source data and driver gaze data
To analyze the driving behavior, this paper proposes a multi-information fusion and visualization, which contains three main modules: data acquisition, data fusion, and feature visualization

Summary

Introduction

According to the technical survey, unsafe driving behavior is the main factor in today’s traffic accidents [1], [2]. Identifying abnormal driving behavior and providing a meaningful evaluation of the driving risk level could help drivers revise their driving behavior and improve the driving safety. Based on the identification and evaluation results, systems like advanced driving assistance systems (ADAS) alert and guide the driver to avoid dangerous driving in time, such as fatigued driving, drunk driving and distracted driving [3]. The evaluation report is beneficial for reasonable traffic signs assignment and user-friendly design of the in-vehicle layout [4]. Drivers have different driving characteristics due to driving skills, driving preferences.

Methods

Results

Conclusion