Traffic Analysis and Safety Enhancement through Deep Learning Models

굴람 무스타파 선문대학교 일반대학원 2024 국내석사

Traffic Analysis and Safety Enhancement through Deep Learning Models This thesis explores the use of deep learning models to improve urban transportation systems, addressing issues like traffic congestion, travel times, productivity reduction, and reduced quality of life. It consists of three interconnected studies, each contributing to a better understanding of urban transportation dynamics and road safety enhance- ment. The first study, conducted in Cheonan city, South Korea, focuses on bus travel time prediction using digital tachograph data from intracity buses. The research evaluates six deep learning models, including Long Short-Term Memory (LSTM) and Gated Recur- rent Unit (GRU) architectures, to predict travel times between Namchang Village bus stop and Dongnam-gu Public Health Center. Comparative analyses reveal that LSTM- based models, particularly Stacked-LSTM, exhibit the highest accuracy. The inclusion of weather data enhances prediction accuracy, providing valuable information for transit agencies and policymakers. The study reports a 47.69% reduction in root mean squared error compared to the baseline ARIMA model, offering data for optimizing public trans- portation schedules and efficiency. The second study focuses on the classification of driver behaviors, with applica- tions in enhancing driver safety, preventing traffic accidents, enabling usage-based insur- ance, and optimizing ridesharing services. The Hybrid ConvLSTM with Attention for Driver Behavior Classification (HCLA-DBC) model adeptly distinguishes between nor- mal, drowsy, and aggressive driving behaviors. Experimental assessments on the UAH- DriveSet dataset reveal better performance than the previous research using the same database, with an accuracy rate of 94.12%, precision score of 94.24%, recall score of 94.12%, F1-score of 94.12%, and ROC AUC of 98.83%. These findings offer detailed in- formation regarding the precise recognition of driver behaviors, which in turn contributes to safer roads, more accurate risk assessment, and enhanced driving experiences. The third study analyzes traffic accidents in Seoul, South Korea, employing the ”TrafficNet” hybrid CNN-FNN model. The model proficiently categorizes accidents into groups, including minor injuries, slander, fatalities, and injury reports. Experiments conducted on traffic accident data from the Seoul Metropolitan Government demonstrate the model’s effectiveness, with an accuracy rate of 93.98%, precision score of 94.31%, recall score of 93.98%, and F1-score of 93.89%. These findings shed light on the primary determinants of traffic accident severity in Seoul, with a specific focus on the influence of vehicle-related factors. This study provides information that can guide the decisions of local authorities and policymakers, with the ultimate goal of improving road safety and mitigating the social and economic consequences of traffic accidents. Keyword: bus travel time prediction; deep learning techniques; digital tachograph data (DTG); Driver behavior classification; Advanced driver-assistance systems (ADAS); Traffic accident prevention; UAH-DriveSet dataset; CNN-FNN hybrid model.

차량 시뮬레이터 활용 운전자 감정에 따른 주행 데이터 분석 및 감정 그룹 제안 연구

이명규 국민대학교 자동차공학전문대학원 2022 국내석사

운전자의 감정이 운전자의 운전 능력에 영향을 미친다는 발표들이 지속적으로 보고되고 있다. 운전자의 감정을 예측하여, 운전 능력을 향상시킬 수 있도록 적절한 서비스를 제공해준다면, 도로 안전은 물론 운전자의 만족도 또한 향상될 것이다. 본 연구는 운전자의 감정에 따라 차량 조작, 생체, 설문조사 데이터에서 어떤 차이를 보이는지 확인하고자 진행되었다. 14명의 실험참가자를 모집하여, 감정을 유도하였고, 차량 시뮬레이터 주행을 요청하였다. 유도하고자 하는 감정으로 행복, 놀람, 두려움, 화남, 슬픔, 지루함, 안도, 중립 여덟 가지를 선정하였다. 여덟 가지 감정을 감정의 각성도와 유인성을 고려하여 Russell의 circumplex model에 배치하였고, 감정을 분류하였다. 감정 유도는 영상 시청, 자기 경험 기술, 차량 시나리오 주행을 통해 진행되었으며, 차량 주행 5분간 실험참가자의 차량 조작, 생체 데이터가, 차량 주행 후 설문조사 데이터가 취득되었다. Human-in-the-loop 실험을 통해, 실험참가자 1명당 하루에 2개의 감정씩 4회, 14명 실험참가자에 대하여 총 56회의 본 실험을 진행하였다. 그 결과, 8개 각각 감정에 따른 차량 조작 데이터, 생체 데이터, 설문조사 데이터를 정상적으로 취득하였다. 결론적으로 14명의 실험참가자를 대상으로 유도하려는 감정을 의도한대로 유도하였고, 각 감정에 따른 데이터를 확보하였다. 통계분석을 통해 감정에 따라 유의한 데이터를 확인할 수 있었고, 유의한 데이터를 바탕으로 8개의 감정을 3개의 그룹으로 분류할 수 있었다. 본 연구 결과는 감정 예측에 대한 기초 연구, 차량 UX design을 위한 파라미터 등으로 활용될 것으로 기대된다. It has been continuously reported that the driver’s emotions affect the driver’s driving ability. If appropriate services are provided to improve driving ability by predicting the driver's emotions in advance, road safety and driver satisfaction will increase. This study was conducted to identify the differences in vehicle control data, physiological data, and survey data according to the driver's emotions. Fourteen experimental participants were recruited and eight emotions were induced. Happiness, surprise, fear, angry, depressed, bored, relieved, and neutral were selected as emotions to induce. Eight emotions were placed in Russell's Circumplex model in consideration of the arousal and valence of emotions, and emotions were classified. Emotional induction was conducted through video watching, writing passage, and scenario driving. As a result, vehicle control data, physiological data, and survey data were obtained after driving for 5 minutes. Through the human-in-the-loop experiment, eight emotions were induced successfully, and vehicle control data, physiological data, and survey data were acquired according to emotions. In conclusion, we were able to accurately induce the emotions, and it was possible to secure data according to each corresponding emotion. Significant data could be confirmed according to emotions, and 8 emotions could be classified into 3 groups based on the significant data. The results of this study can be used as a basic study on emotion prediction and parameters for vehicle UX design.

Analysis of the Impact Factor on Road Traffic Accidents under Introduction of Road Safety Countermeasures in Lao PDR

BOUNSOMBATH, Chanthavong University of Seoul Internatinal School of Urban S 2021 국내석사

도로 교통 사고(Road traffic accident, RTA)와 부상은 세계적으로 중요한 보건 문제다. 대부분의 사고는 인간 행동, 도로 상태 및 차량이라는 세 가지 주요 인자로 인해 발생한다. 본 연구의 목적은 라오스에서 도로 교통 사고의 주요 인자를 분석하고, 중요한 문제가 뭔지 파악하며, 적절한 안전 대책을 제안하는 것이다. 본 연구에서는 분석, 파악 및 비교에 초점을 맞추어 RTA 원인 제공을 설명하기로 한다. 뿐만 아니라 2010~2019년 라오스에서 발생한 도로 교통 사고 통계 자료(이차 자료)를 이용했다. 라오스에서 도로 교통 사고를 유발하는 주요 기여 인자는 무엇인가? 바로 인간 행동, 음주 운전, 과속 및 교통 위반인 것으로 나타났다. 더불어 도로 교통 사고를 유발한 차량 유형과 도로 상태 유형도 기여 인자로 나타났다. 데이터 분석과 RTA 통계 결과 세 가지 주요 인자가 나왔는데 바로 무면허 운전자의 교통 위반, 음주 운전 및 과속이었다. 마지막으로 본 연구는 미래 도로 사고 수를 줄이고 예방하기 위한 도로 안전 대책에 관한 권고와 지침을 제안했다. Road traffic accidents (RTAs) and injuries are the major global public health problem. Most accidents occur due to three main factors, such as human behaviors, road conditions, and vehicles. This study aims to analyze the main factors of road traffic accidents, identify the critical issues, and suggest appropriate safety countermeasures in Lao PDR. This research will explain the contribution of RTAs by focusing on analyses, identification, and comparison. Furthermore, this study will be conducted road traffic accident statistic data (secondary data) with the time frame from 2010-2019 in Lao PDR. What are the main contributing factors to road traffic accidents in Lao PDR? Factors such as human behaviors, drink-drive, over speed, traffic offenses. Moreover, type of vehicles involved in road traffic accidents and type of road condition. Three main factors are derived from data analyses and RTAs statistics. These are traffic offense by unlicensed of driver, drink drive and over speed. Finally, this thesis suggests recommendations or guidelines for road safety countermeasures to reduce and prevention the number of road accidents for the future.

Factors Associated with Obstructive Sleep Apnea Evaluation in At-Risk Patients Generally and in the Perianesthesia Setting Specifically

Orbell, Staci L University of Pittsburgh ProQuest Dissertations & 2023 해외박사(DDOD)

Background/Purpose:Obstructive sleep apnea (OSA) is a highly prevalent yet underdiagnosed sleep-related breathing disorder. While studies have been conducted to examine factors associated with OSA care-seeking in at-risk individuals, it is unclear which factors are most influential. Further, these factors have not been explored in at-risk patients identified in the perianesthesia setting, in spite of this care specialty’s provision of routine OSA screening. We aimed to address these gaps by reviewing current literature on factors associated with OSA evaluation overall, and in patients identified as at-risk for OSA in the perianesthesia setting, examining associations between OSA care-seeking behavior and health related factors overall, and by age, sex, and marital status.Methods:A mixed methods literature review was performed to examine factors associated with OSA evaluation. Eligible articles addressed patient, provider, or system-level factors impacting completion of an OSA diagnostic evaluation, care-seeking and/or adherence rates. An observational study was also conducted in a sample of at-risk adults who received OSA risk notification and recommendation for follow-up evaluation as part of an outpatient procedure. Logistic regression examined associations between adherence to a provider’s recommendation for OSA evaluation and demographic, clinical and health-related factors. Linear regression examined these same factors and associations between OSA care-seeking intention stratified by age, sex, and marital status.Results/Conclusion:Twenty-six articles including quantitative, qualitative, and mixed methods studies were included in the literature review. Factors found to be most influential to OSA careseeking and/or evaluation were social support, sex and the influence of gender, OSA-related symptoms and experiences, OSA knowledge and beliefs, healthcare provider involvement, and administrative considerations. In the original research arm of this study, in a sample of 63 patients identified as at-risk for OSA in the perianesthesia setting, 12.7% adhered to a provider’s recommendation for follow-up evaluation. Excessive daytime sleepiness was identified as the strongest predictor of follow-up adherence. Functional impairment related to sleepiness and perceived likelihood of having OSA were the strongest predictors of OSA care-seeking intention. Functional impairment was important to OSA care-seeking intention in younger adults and regardless of sex or marital status; perceived likelihood of having OSA was an important predictor in men.