Vehicular Trajectory Big Data: Driving Behavior Recognition Algorithm Based on Deep Learning

Yang, Xinghai; Ding, Fei; Zhang, Dengyin; Zhang, Min

doi:10.1007/978-981-15-8086-4_30

Cited by 5 publications

(4 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, the CP system using random forest was able to predict CS and NCS with high accuracy based on θ ( y ) avg and ω ( x ) avg of an older driver while preparing to start the car. Related works predicted driver risk [ 33 ], typical dangerous driving behavior such as chasing a preceding vehicle [ 34 ] or driving while operating a mobile phone [ 35 ], and illegal drivers [ 36 ]. However, there is still not a prediction about the coping skills of older drivers in the face of unexpected situations; although, it is not possible to make a direct comparison because the prediction targets are different, the related study predicts dangerous driving behavior with an accuracy of 92% [ 34 ], and the accuracy of this study is about the same even when compared with related work.…”

Section: Discussionmentioning

confidence: 99%

Predicting the Coping Skills of Older Drivers in the Face of Unexpected Situation

Kajiwara

Kimura

2021

Sensors

View full text Add to dashboard Cite

In recent years, when an older driver who cannot immediately recognize, judge, and operate properly faces an unexpected situation, they often panic, which may cause a traffic accident. However, there has not yet been enough discussion about the coping skills of older drivers in the face of this unexpected situation. Therefore, this study discusses the coping skills of older drivers in the face of unexpected situations. Moreover, we propose a coping skills prediction system (CP system). The CP system predicts coping skills from the tilt angle and angular velocity of the left foot when an older driver is driving or preparing to start a car. The experiment carried out two phases, a phase of driving a car and a phase of preparing to start the car. In the driving phase, the young and older driver drive the car in a driving simulator. The average age of the young driver group was ± standard deviation = 20.6 ± 0.7 years, and the age of the older driver group was 78.5 ± 5.1 years. The driving route included 15 cases in which collision accidents are likely to occur. We analyzed the experimental results of the driving phase and clarified the predictors of coping skills. Moreover, we analyzed the correlation between the left foot movement in driving and the left foot movement during preparing to start the car. As a result of the experiment, there was a 0.84 correlation between the tilt angle of the left foot of the older driver in driving and the tilt angle of the left foot of the older driver in preparing to start the car. The result shows that the coping skills can be predicted from the tilt angle of the left foot of the older driver during preparing to start the car. We showed that the coping skill can be predicted with an accuracy of 92% or 94% on average from the tilt angle and the angular velocity of the left foot while driving or preparing to start the car. Moreover, we clarified that the tilt angle of the left foot of a driver without coping skills is perpendicular to the ground compared to a driver with coping skills. This study is expected to contribute to the prevention of traffic accidents that occur in the face of an unexpected situation.

show abstract

Section: Discussionmentioning

confidence: 99%

Predicting the Coping Skills of Older Drivers in the Face of Unexpected Situation

Kajiwara

Kimura

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…This amalgamation facilitates face detection achieved through feature filtering employing a cascaded classifier powered by the Ada Boost algorithm [1]. Shan Zhang, on the other hand, introduces a motion recognition scheme founded on facial analysis, spanning from a comprehensive view to localized scrutiny [2]. Dewei Zheng, in his work, extracts fatigue features by utilizing an optimal set of facial feature points and subsequently employs statistical analysis methods to scrutinize and validate the efficacy of these features [3].…”

Section: Introductionmentioning

confidence: 99%

Intelligent Recognition and Detection Method for Facial Posture of Motor Vehicle Drivers Based on Computer Vision

Cao

2024

Advances in Transdisciplinary Engineering

View full text Add to dashboard Cite

The driver monitoring system constitutes an essential element in the realm of human-machine interaction within intelligent vehicles, primarily tasked with overseeing and promptly alerting deviations from standard driving behaviors that could potentially lead to traffic accidents. Presently, the evolution of driver monitoring systems in China is at a nascent stage. Challenges persist due to constraints in hardware equipment, resulting in relatively simplistic terminal devices for detecting driver fatigue features. Consequently, this has led to frequent occurrences of false positives and missed detections. This study centers around a non-contact vehicle monitoring device employing machine vision. It delves into the development of a rational, effective, real-time, and accurate mechanism for continuously monitoring driving duration while concurrently recording driving behavior data. Within this framework, the research focuses on the utilization of a modified version of MTCNN for real-time driver face detection. It involves extracting critical driver head posture features, encompassing the pitch angle, yaw angle, and roll angle of the driver’s head. By comparing the positions of obtained driver face images, a fatigue driving feature index system is established. This system facilitates the identification, analysis, and discrimination of a driver’s fatigue state based on the extracted head feature values. The ultimate aim is to realize a fatigue warning function within the driver monitoring system, thereby enhancing the detection speed of facial fatigue recognition. This endeavor holds paramount significance for road traffic safety, contributing to the continual improvement of driver monitoring systems and consequently mitigating potential risks on the road.

show abstract

“…Vehicle trajectory extraction, reconstruction, filtering, or smoothing are closely related and have attracted research interest more than half a century ago with the seminal works of 1 (i.e., Wiener filter), and 2 (i.e., Rauch-Tung-Striebel smoother). It is an important topic that can enable investigations on various other topics, e.g., emissions or energy demand 3 , 4 , traffic dynamics 5 – 8 , modeling and control 9 – 11 , control 11 , connected and automated vehicles and driver understanding 12 , 13 , while this list is not exhaustive.…”

Section: Introductionmentioning

confidence: 99%

“…Different data acquisition devices employ different characteristics (accuracy, noise, sampling frequency), and raw data, in most cases, if not all, need post-processing.Vehicle trajectory extraction, reconstruction, filtering, or smoothing are closely related and have attracted research interest more than half a century ago with the seminal works of 1 (i.e., Wiener filter), and 2 (i.e., Rauch-Tung-Striebel smoother). It is an important topic that can enable investigations on various other topics, e.g., emissions or energy demand 3,4 , traffic dynamics 5-8 , modeling and control 9-11 , control 11 , connected and automated vehicles and driver understanding 12,13 , while this list is not exhaustive.At the same time, more and more datasets with experimental observations have become publicly available, and nowadays, it becomes easier to organize experimental campaigns and study complex phenomena empirically. Filtering techniques are applied to the raw data to remove noise and obtain quality measurements.…”

mentioning

confidence: 99%

Adaptive physics-informed trajectory reconstruction exploiting driver behavior and car dynamics

Makridis

Kouvelas

2023

Sci Rep

View full text Add to dashboard Cite

As more and more trajectory data become available, their analysis creates unprecedented opportunities for traffic flow investigations. However, observed physical quantities like speed or acceleration are often measured having unrealistic values. Furthermore, observation devices have different hardware and software specifications leading to heterogeneity in noise levels and limiting the efficiency of trajectory reconstruction methods. Typical strategies prune, smooth, or locally modify vehicle trajectories to infer physically plausible quantities. The filtering strength is usually heuristic. Once the physical quantities reach plausible values, additional improvement is impossible without ground truth data. This paper proposes an adaptive physics-informed trajectory reconstruction framework that iteratively detects the optimal filtering magnitude, minimizing local acceleration variance under stable conditions and ensuring compatibility with feasible vehicle acceleration dynamics and common driver behavior characteristics. Assessment is performed using both synthetic and real-world data. Results show a significant reduction in the speed error and invariability of the framework to different data acquisition devices. The last contribution enables the objective comparison between drivers with different sensing equipment.

show abstract

Vehicular Trajectory Big Data: Driving Behavior Recognition Algorithm Based on Deep Learning

Cited by 5 publications

References 19 publications

Predicting the Coping Skills of Older Drivers in the Face of Unexpected Situation

Predicting the Coping Skills of Older Drivers in the Face of Unexpected Situation

Intelligent Recognition and Detection Method for Facial Posture of Motor Vehicle Drivers Based on Computer Vision

Adaptive physics-informed trajectory reconstruction exploiting driver behavior and car dynamics

Contact Info

Product

Resources

About