Risky Driver Recognition with Class Imbalance Data and Automated Machine Learning Framework

Wang, Ke; Xue, Qianqian; Lu, Jian

doi:10.3390/ijerph18147534

Cited by 11 publications

(4 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model verification was performed through HIL simulation, achieving 83.75% prediction accuracy. Wang et al [220] developed a risky driver prediction model using video data from an overhead traffic drone. The drivers from the video are labeled according to their average collision risk, which is calculated based on the spacing and stopping distance (DSS).…”

Section: A Human Driver Behavior Recognitionmentioning

confidence: 99%

Driver Behavior Modeling Toward Autonomous Vehicles: Comprehensive Review

Negash

Yang

2023

IEEE Access

View full text Add to dashboard Cite

Driver behavior models have been used as input to self-coaching, accident prevention studies, and developing driver-assisting systems. In recent years, driver behavior recognition has revolutionized autonomous vehicles (AVs) and traffic management studies. This comprehensive survey provides an up-todate review of the different driver behavior models and modeling approaches. In heterogeneous streets where humans and autonomous vehicles operate simultaneously, predicting the intent and action of human drivers is crucial for AVs with the help of wireless communication and artificial intelligence (AI) technologies. Therefore, the review also summarizes the applications of driver behavior modeling (DBM) for effective behavior recognition and human-like AV driving. Moreover, the review also covers the application of DBM in capturing behaviors of complex dynamic driving tasks. In this review, we solely cover car-following (CF) and lane-changing (LC) maneuvers.INDEX TERMS Driver behavior modeling (DBM), autonomous vehicles, behavior recognition, human-like, car-following, lane-changing, perception, decision-making, data-driven, vehicle control. I. INTRODUCTIONIn 2018, the US Department of Transportation reported 6,734,000 motor vehicle-related accidents. 1 Among these accidents, there were 2.71 million injuries and 36,560 fatalities. These deaths and injuries also increased in the years 2019 and 2020. Earlier in 2015, specifically in the US, 80 − 94% of crashes were due to driver-related errors, and 33% were due to the wrong prediction of other drivers [1]. The statistics of accidents caused by human driver-related errors were also reported in [2] and [3].The accidents and their behind reasons will lead us back to human drivers with rich and diverse behaviors. According to Michon [4], a human as a problem solver can be a road user, a transportation consumer, a social agent, and a psycho-biological organism. The issues to be solved are controlling a vehicle, trip-making, communication, and satisfying the basic driving necessity. A human driver can have unique traits towards vehicle controlThe associate editor coordinating the review of this manuscript and approving it for publication was Divanilson Rodrigo Campelo .

show abstract

Section: A Human Driver Behavior Recognitionmentioning

confidence: 99%

Driver Behavior Modeling Toward Autonomous Vehicles: Comprehensive Review

Negash

Yang

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The ratio of the absolute value of DSS to the speed of the rear vehicle can reflect the missing reaction time required by the driver [26]. The formula is shown in Equation (6).…”

Section: Lane-changing Risk Indicatormentioning

confidence: 99%

Driving Behavior Risk Measurement and Cluster Analysis Driven by Vehicle Trajectory Data

Chen

Cheng²,

Yang

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

The correct identification and timely pre-warning of driving behavior risks can remind drivers to correct their unsafe driving behaviors effectively. First of all, four risk evaluation indicators of driving behavior were defined based on lateral and longitudinal driving characteristics: the lateral stability indicator, the longitudinal stability indicator, the car-following risk indicator, and the lane-changing risk indicator. The Pearson correlation coefficient method was used to analyze the correlation of the four indicators, and the conclusion showed that the four indicators were very weakly correlated or presented an irrelevant correlation. Thus, the four indicators can describe different driving behavior risks. Secondly, the criteria importance through intercriteria correlation (CRITIC) method was used to determine the weight of each indicator, and a comprehensive measurement model of driving behavior risk was established. To test the model, this study preprocessed the trajectory data of small vehicles in Lanes 1–5 of the I-80 Expressway from the NGSIM dataset, collected statistical analysis results of vehicle speed and acceleration, and obtained the parameters data required for risk assessment. Then, based on the obtained trajectory data, the variation laws and the thresholds of the four indicators were determined by using the interquartile difference method. Finally, by using the K-means clustering algorithm, the risk types of driving behavior were divided into four categories, namely, dangerous, aggressive, safe, and conservative. The dangerous, aggressive, safe, and conservative driving behaviors accounted for 5.40%, 23.30%, 43.22%, and 28.08% of the total samples, respectively. The expert’s assessment results of the driving behavior risk aligned with the results obtained from the model measurements. This indicated that the driving behavior risk measurement model here described can evaluate a driver’s risk status in real time, provide safety tips for the driver, and offer theoretical support for driving safety warning systems.

show abstract

“…The most common sampling techniques in the literature are the Synthetic Minority Oversampling Technique (SMOTE) [ 16 , 22 , 23 , 24 ] and Adaptive Synthetic (ADASYN) [ 24 ]. In addition, based on the literature review in the field of road safety as well as different scientific areas, additional sampling techniques tend to be efficient methods such as the combination of SMOTE and Edited Nearest Neighbors (SMOTE-ENN) [ 10 ], Random Oversampling, SVM-SMOTE and SMOTE-Tomek [ 25 ].…”

Section: Literature Reviewmentioning

confidence: 99%

“…In summary, this research proposes a framework for (a) defining the STZ levels, and (b) developing and evaluating machine learning algorithms to classify driving behavior and predict the duration that each driver spends at each risk level. This framework also exploits the most important features to identify driving behavior and takes care of dataset imbalance, which is a common problem in road safety analyses [ 10 ]. The paper contributes to the current knowledge in a two-fold matter; initially by identifying the level of safety of drivers in real-time, which is a real-time classification problem, and consequently by predicting the duration of each safety level in real-time.…”

Section: Introductionmentioning

confidence: 99%

Data-Driven Estimation of a Driving Safety Tolerance Zone Using Imbalanced Machine Learning

Garefalakis

Katrakazas

Yannis

2022

Sensors

View full text Add to dashboard Cite

Predicting driving behavior and crash risk in real-time is a problem that has been heavily researched in the past years. Although in-vehicle interventions and gamification features in post-trip dashboards have emerged, the connection between real-time driving behavior prediction and the triggering of such interventions is yet to be realized. This is the focus of the European Horizon2020 project “i-DREAMS”, which aims at defining, developing, testing and validating a ‘Safety Tolerance Zone’ (STZ) in order to prevent drivers from risky driving behaviors using interventions both in real-time and post-trip. However, the data-driven conceptualization of STZ levels is a challenging task, and data class imbalance might hinder this process. Following the project principles and taking the aforementioned challenges into consideration, this paper proposes a framework to identify the level of risky driving behavior as well as the duration of the time spent in each risk level by private car drivers. This aim is accomplished by four classification algorithms, namely Support Vector Machines (SVMs), Random Forest (RFs), AdaBoost, and Multilayer Perceptron (MLP) Neural Networks and imbalanced learning using the Adaptive Synthetic technique (ADASYN) in order to deal with the unbalanced distribution of the dataset in the STZ levels. Moreover, as an alternative approach of risk prediction, three regression algorithms, namely Ridge, Lasso, and Elastic Net are used to predict time duration. The results showed that RF and MLP outperformed the rest of the classifiers with 84% and 82% overall accuracy, respectively, and that the maximum speed of the vehicle during a 30 s interval, is the most crucial predictor for identifying the driving time at each safety level.

show abstract

Risky Driver Recognition with Class Imbalance Data and Automated Machine Learning Framework

Cited by 11 publications

References 43 publications

Driver Behavior Modeling Toward Autonomous Vehicles: Comprehensive Review

Driver Behavior Modeling Toward Autonomous Vehicles: Comprehensive Review

Driving Behavior Risk Measurement and Cluster Analysis Driven by Vehicle Trajectory Data

Data-Driven Estimation of a Driving Safety Tolerance Zone Using Imbalanced Machine Learning

Contact Info

Product

Resources

About