scholarly journals A Crash Prediction Method Based on Artificial Intelligence Techniques and Driving Behavior Event Data

2021 ◽  
Vol 13 (11) ◽  
pp. 6102
Author(s):  
Yunjong Kim ◽  
Juneyoung Park ◽  
Cheol Oh
Keyword(s):  
Traffic Flow ◽  
Traffic Accidents ◽  
Prediction Method ◽  
Driving Behavior ◽  
Crash Risk ◽  
Gradient Boosting ◽  
Event Data ◽  
Trajectory Data ◽  
Data Set ◽  
Ensemble Technique

Various studies on how to prevent and deal with traffic accidents are ongoing. In the past, the key research emphasis was on passive accident response measures that analyzed roadway-based historical data to identify road sections with high crash risk. Through assessing crash risks by analyzing simulation data and actual vehicle driving trajectory data, this study suggests a method of effectively preventing accidents before they happen. In this analysis, using digital tachograph (DTG) data, which is the vehicle trajectory data for commercial vehicles running on Korean highways, hazardous and normal traffic flows were identified and extracted. Driving behavior event data for both types of traffic flow was processed by measuring safety indicators through the extracted data. Safety indicators with a high impact on traffic flow classification were then extracted using gradient boosting, a representative ensemble technique. A neural network analysis was performed using the extracted safety indicators as independent variables to create a traffic flow classifier, which had a high accuracy of 94.59%. The DTG data set was also classified based on the severity of each accident that occurred in the studied roadway, the time of the accident, and the weather; the results were compiled to enable comprehensive accident prediction. It is expected that proactive crash prevention will be possible in the future by evaluating real-time accident risks using the findings and ensemble-based methodologies of this paper.

Driving Behavior at Signalized Intersections Operating under Disordered Traffic Conditions

2021 ◽  
pp. 036119812110338
Author(s):  
Ritvik Chauhan ◽  
Ashish Dhamaniya ◽  
Shriniwas Arkatkar
Keyword(s):  
Traffic Flow ◽  
Flow Analysis ◽  
Driving Behavior ◽  
Signalized Intersections ◽  
Lateral Movement ◽  
Trajectory Data ◽  
Lateral Velocity ◽  
Data Set ◽  
Vehicle Class ◽  
Traffic Conditions

A higher degree of heterogeneity in vehicle class and drivers, coupled with non-lane-based driving habits, creates several challenges in traffic flow analysis. This study investigates vehicles’ microscopic driving behavior at signalized intersections operating under weak lane discipline with mixed traffic (disordered) conditions. For this purpose, a comprehensive vehicular trajectory data set is developed from field-recorded video footage using a semi-automated tool for data extraction. Microscopic parameters such as relative velocity, spacing between vehicles, following time, lane preference, longitudinal and lateral speed profile, hysteresis evidence, and lateral movement of different vehicle classes during different traffic phases are presented in the study. The data is then segregated into three flow conditions: stopped flow, saturated flow, and unaffected flow. It is found that smaller vehicles prefer near-side lanes over far-side lanes. Motorized three-wheeler (3W) and motorized two-wheeler (2W) vehicle classes exhibit the greatest lateral velocity, lateral movement, and aggressiveness. This results in several interactions between vehicles as a function of different leader–follower vehicle pairs. Signalized intersections with more heterogeneity in traffic composition, especially higher composition of 2W and 3W vehicle classes, exhibit higher levels of aggressive driving behavior that might lower safety standards. As a practical application, ranges of various driving behavior parameter values for different leader–follower combinations and traffic conditions are quantified in the study. The observations and results are expected to help better understand prevailing driving behavior in disordered traffic and contribute toward robust calibration of microscopic traffic flow models for better replicating disordered traffic conditions at signalized intersections.

scholarly journals Research on Mandatory Lane-Changing Behavior in Highway Weaving Sections

2020 ◽  
Vol 2020 ◽  
pp. 1-9 ◽  
Author(s):  
Wei Hao ◽  
Zhaolei Zhang ◽  
Zhibo Gao ◽  
Kefu Yi ◽  
Li Liu ◽  
...  
Keyword(s):  
Traffic Flow ◽  
Traffic Accidents ◽  
Driving Behavior ◽  
Heterogeneous Traffic ◽  
Psychological Characteristics ◽  
Distribution Characteristics ◽  
Mixed Traffic ◽  
Lane Changing ◽  
Headway Distribution ◽  
Analysis Models

As the accident-prone sections and bottlenecks, highway weaving sections will become more complicated when it comes to the mixed-traffic environments with connected and automated vehicles (CAVs) and human-driven vehicles (HVs). In order to make CAVs accurately identify the driving behavior of manual-human vehicles to avoid traffic accidents caused by lane changing, it is necessary to analyze the characteristics of the mandatory lane-changing (MCL) process in the weaving area. An analytical MCL method based on the driver’s psychological characteristics is proposed in this study. Firstly, the driver’s MLC pressure concept was proposed by leading in the distance of the off-ramp. Then, the lane-changing intention was quantified by considering the driver’s MLC pressure and tendentiousness. Finally, based on the lane-changing intention and the headway distribution of the target lane, an MLC positions probability density model was proposed to describe the distribution characteristics of the lane-changing position. Through the NGSIM data verification, the lane-changing analysis models can objectively describe the vehicle lane-changing characteristics in the actual scenarios. Compared with the traditional lane-changing model, the proposed models are more interpretable and in line with the driving intention. The results show significant improvements in the lane-changing safe recognition of CAVs in heterogeneous traffic flow (both CAVs and HVs) in the future.

scholarly journals Comparison of algorithms for road surface temperature prediction

2018 ◽  
Vol 2 (3) ◽  
pp. 212-224
Author(s):  
Bo Liu ◽  
Libin Shen ◽  
Huanling You ◽  
Yan Dong ◽  
Jianqiang Li ◽  
...  
Keyword(s):  
Surface Temperature ◽  
Prediction Method ◽  
Temperature Data ◽  
Road Surface ◽  
Machine Learning Algorithms ◽  
Gradient Boosting ◽  
Data Set ◽  
Content Type ◽  
The Road ◽  
Road Surface Temperature

Purpose The influence of road surface temperature (RST) on vehicles is becoming more and more obvious. Accurate predication of RST is distinctly meaningful. At present, however, the prediction accuracy of RST is not satisfied with physical methods or statistical learning methods. To find an effective prediction method, this paper selects five representative algorithms to predict the road surface temperature separately. Design/methodology/approach Multiple linear regressions, least absolute shrinkage and selection operator, random forest and gradient boosting regression tree (GBRT) and neural network are chosen to be representative predictors. Findings The experimental results show that for temperature data set of this experiment, the prediction effect of GBRT in the ensemble algorithm is the best compared with the other four algorithms. Originality/value This paper compares different kinds of machine learning algorithms, observes the road surface temperature data from different angles, and finds the most suitable prediction method.

Investigating Effect of Driver-, Vehicle-, and Road-Related Factors on Location-Specific Crashes with Naturalistic Driving Data

2019 ◽  
Vol 2673 (6) ◽  
pp. 46-56 ◽  
Author(s):  
Grace Ashley ◽  
Osama A. Osman ◽  
Sherif Ishak ◽  
Julius Codjoe
Keyword(s):  
United States ◽  
Machine Learning ◽  
Traffic Accidents ◽  
The United States ◽  
Crash Risk ◽  
Related Factors ◽  
Data Set ◽  
Intersection Safety ◽  
Naturalistic Driving Study ◽  
Naturalistic Driving

According to NHTSA, traffic accidents cost the United States billions of U.S. dollars each year. Intersection accidents alone accounted for 23% of the 32,675 motor crash deaths in 2014. With the advent of the largest naturalistic driving data set in the United States collected by the SHRP2 Naturalistic Driving Study project, this study performs a crash-only analysis to identify driver-, vehicle-, and roadway-related factors that affect the driving risk at different location types using a machine learning tool. The study then analyzes the most important factors obtained from the machine learning analysis to identify how they affect crash risk. The results, in order of importance of variables, were driver behavior, locality, lane occupied, alignment, and through travel lanes. Also, drivers who violated traffic signals were four times more likely to be involved in a crash than drivers who did not. Those who violated stop signs were two times more likely to be involved in crashes than those who did not. Drivers performing visual-manual (VM) tasks at uncontrolled intersections were 2.7 times more likely to be involved in crashes than those who did not engage in these tasks. At nonintersections, drivers who performed VM tasks were 3.4 times more likely to be involved in crashes than drivers who did not. These findings add to the evidence that the establishment of safety awareness programs geared toward intersection safety is imperative.

A study of freeway crash risk prediction and interpretation based on risky driving behavior and traffic flow data

2021 ◽  
Vol 160 ◽  
pp. 106328
Author(s):  
Miao Guo ◽  
Xiaohua Zhao ◽  
Ying Yao ◽  
Pengwei Yan ◽  
Yuelong Su ◽  
...  
Keyword(s):  
Traffic Flow ◽  
Risk Prediction ◽  
Driving Behavior ◽  
Crash Risk ◽  
Flow Data ◽  
Risky Driving

scholarly journals Deep Reinforcement Learning Based on Link Prediction Method in Social Network Analysis

2019 ◽  
Vol 9 (2S2) ◽  
pp. 820-826
Keyword(s):  
Machine Learning ◽  
Social Network ◽  
Reinforcement Learning ◽  
Supervised Learning ◽  
Link Prediction ◽  
Prediction Models ◽  
Learning Strategy ◽  
Prediction Method ◽  
Gradient Boosting ◽  
Data Set

Improving the performance of link prediction is a significant role in the evaluation of social network. Link prediction is known as one of the primary purposes for recommended systems, bio information, and web. Most machine learning methods that depend on SNA model’s metrics use supervised learning to develop link prediction models. Supervised learning actually needed huge amount of data set to train the model of link prediction to obtain an optimal level of performance. In few years, Deep Reinforcement Learning (DRL) has achieved excellent success in various domain such as SNA. In this paper, we present the use of deep reinforcement learning (DRL) to improve the performance and accuracy of the model for the applied dataset. The experiment shows that the dataset created by the DRL model through self-play or auto-simulation can be utilized to improve the link prediction model. We have used three different datasets: JUNANES, MAMBO, JAKE. Experimental results show that the DRL proposed method provide accuracy of 85% for JUNANES, 87% for MAMABO, and 78% for JAKE dataset which outperforms the GBM next highest accuracy of 75% for JUNANES, 79% for MAMBO and 71% for JAKE dataset respectively trained with 2500 iteration and also in terms of AUC measures as well. The DRL model shows the better efficiency than a traditional machine learning strategy, such as, Random Forest and the gradient boosting machine (GBM).

scholarly journals Early Prediction of Refractory Epilepsy in Children Under Artificial Intelligence Neural Network

2021 ◽  
Vol 15 ◽  
Author(s):  
Yueyan Huang ◽  
Qingfeng Li ◽  
Qian Yang ◽  
Zhijing Huang ◽  
Hongbo Gao ◽  
...  
Keyword(s):  
Neural Network ◽  
Refractory Epilepsy ◽  
Back Propagation ◽  
Prediction Method ◽  
Gradient Boosting ◽  
Support Vector ◽  
Early Prediction ◽  
Data Set ◽  
Adaboost Algorithm ◽  
Public Data

In order to realize the early prediction of refractory epilepsy in children, data preprocessing technology was used to improve the data quality, and the detection model of refractory epilepsy in children based on convolutional neural network (CNN) was established. Then, the data in the epilepsy electroencephalography (EEG) signal public data set was used for model training and the diagnosis of refractory epilepsy in children. Moreover, back propagation neural network (BPNN), support vector machine (SVM), XGBoost, gradient boosting decision tree (GBDT), AdaBoost algorithm were introduced for comparison. The results showed that the early prediction accuracy of BP, SVM, XGBoost, GBDT, AdaBoost, and the algorithm in this study for refractory epilepsy in children were 0.745, 0.778, 0.885, 0.846, 0.874, and 0.941, respectively. The sensitivities were 0.81, 0.826, 0.822, 0.84, 0.859, and 0.918, respectively. The specificities were 0.683, 0.696, 0.743, 0.792, 0.84, and 0.905, respectively. The accuracy was 0.707, 0.732, 0.765, 0.802, 0.839, and 0.881, respectively. The recall rates were 0.69, 0.716, 0.753, 0.784, 0.813, and 0.877, respectively. F1 scores were 0.698, 0.724, 0.759, 0.793, 0.826, and 0.879, respectively. Through the comparisons of the above six indicators, the algorithm proposed in this study was significantly higher than other algorithms, suggesting that the proposed algorithm was more accurate in early prediction of refractory epilepsy in children. Analysis of the EEG characteristics and magnetic resonance imaging (MRI) images of refractory epilepsy in children suggested that the MRI images of patients' brains under this algorithm had obvious characteristics. The reason for the prediction error of the algorithm was that the duration of epilepsy was too short or the EEG of the patient didn't change notably during the epileptic seizure. In summary, the prediction method of refractory epilepsy in children based on CNN was accurate, which had broad adoption prospects in assisting clinicians in the examination and diagnosis of refractory epilepsy in children.

Application of Trajectory Data for Investigating Vehicle Behavior in Mixed Traffic Environment

2018 ◽  
Vol 2672 (43) ◽  
pp. 122-133 ◽  
Author(s):  
Narayana Raju ◽  
Pallav Kumar ◽  
Aayush Jain ◽  
Shriniwas S. Arkatkar ◽  
Gaurang Joshi
Keyword(s):  
Traffic Flow ◽  
Research Work ◽  
Driving Behavior ◽  
Video Data ◽  
Hysteresis Phenomenon ◽  
Mixed Traffic ◽  
Trajectory Data ◽  
Traffic Conditions ◽  
Micro Level ◽  
Flow States

The research work reported here investigates driving behavior under mixed traffic conditions on high-speed, multilane highways. With the involvement of multiple vehicle classes, high-resolution trajectory data is necessary for exploring vehicle-following, lateral movement, and seeping behavior under varying traffic flow states. An access-controlled, mid-block road section was selected for video data collection under varying traffic flow conditions. Using a semi-automated image processing tool, vehicular trajectory data was developed for three different traffic states. Micro-level behavior such as lateral placement of vehicles as a function of speed, instant responses, vehicle-following behavior, and hysteresis phenomenon were evaluated under different traffic flow states. It was found that lane-wise behavior degraded with increase in traffic volume and vehicles showed a propensity to move towards the median at low flow and towards the curb-side at moderate and heavy flows. Further, vehicle-following behavior was also investigated and it was found that with increase in flow level, vehicles are more inclined to mimic the leader vehicle’s behavior. In addition to following time, perceiving time of subject vehicle for different leading vehicles was also evaluated for different vehicle classes. From the analysis, it was inferred that smaller vehicles are switching their leader vehicles more often to escape from delay, resulting in less following and perceiving time and aggressive gap acceptance. The present research work reveals the need for high-quality, micro-level data for calibrating driving behavior models under mixed traffic conditions.

In silico Prediction of Inhibitory Constant of Thrombin Inhibitors Using Machine Learning

2019 ◽  
Vol 21 (9) ◽  
pp. 662-669 ◽  
Author(s):  
Junnan Zhao ◽  
Lu Zhu ◽  
Weineng Zhou ◽  
Lingfeng Yin ◽  
Yuchen Wang ◽  
...  
Keyword(s):  
Machine Learning ◽  
Prediction Models ◽  
Regression Tree ◽  
Large Data ◽  
Thrombin Inhibitors ◽  
Coagulation Cascade ◽  
Gradient Boosting ◽  
Support Vector ◽  
Data Set ◽  
Descriptor Selection

Background: Thrombin is the central protease of the vertebrate blood coagulation cascade, which is closely related to cardiovascular diseases. The inhibitory constant Ki is the most significant property of thrombin inhibitors. Method: This study was carried out to predict Ki values of thrombin inhibitors based on a large data set by using machine learning methods. Taking advantage of finding non-intuitive regularities on high-dimensional datasets, machine learning can be used to build effective predictive models. A total of 6554 descriptors for each compound were collected and an efficient descriptor selection method was chosen to find the appropriate descriptors. Four different methods including multiple linear regression (MLR), K Nearest Neighbors (KNN), Gradient Boosting Regression Tree (GBRT) and Support Vector Machine (SVM) were implemented to build prediction models with these selected descriptors. Results: The SVM model was the best one among these methods with R2=0.84, MSE=0.55 for the training set and R2=0.83, MSE=0.56 for the test set. Several validation methods such as yrandomization test and applicability domain evaluation, were adopted to assess the robustness and generalization ability of the model. The final model shows excellent stability and predictive ability and can be employed for rapid estimation of the inhibitory constant, which is full of help for designing novel thrombin inhibitors.

Cloning Safe Driving Behavior for Self-Driving Cars using Convolutional Neural Networks

2019 ◽  
Vol 12 (2) ◽  
pp. 120-127 ◽  
Author(s):  
Wael Farag
Keyword(s):  
Gradient Descent ◽  
Autonomous Driving ◽  
Driving Behavior ◽  
Training Data ◽  
Stochastic Gradient Descent ◽  
Data Set ◽  
Safe Driving ◽  
Processing Pipeline ◽  
Self Driving Cars ◽  
And Training

Background: In this paper, a Convolutional Neural Network (CNN) to learn safe driving behavior and smooth steering manoeuvring, is proposed as an empowerment of autonomous driving technologies. The training data is collected from a front-facing camera and the steering commands issued by an experienced driver driving in traffic as well as urban roads. Methods: This data is then used to train the proposed CNN to facilitate what it is called “Behavioral Cloning”. The proposed Behavior Cloning CNN is named as “BCNet”, and its deep seventeen-layer architecture has been selected after extensive trials. The BCNet got trained using Adam’s optimization algorithm as a variant of the Stochastic Gradient Descent (SGD) technique. Results: The paper goes through the development and training process in details and shows the image processing pipeline harnessed in the development. Conclusion: The proposed approach proved successful in cloning the driving behavior embedded in the training data set after extensive simulations.

Sign in / Sign up

Export Citation Format

Share Document