All Issue

2025 Vol.43, Issue 5 Preview Page

Article

Development of Autonomous Vehicle Edge Case Scenarios Based on Topic Modelling Through Dashcam Video Analysis : Focusing on Pedestrian, Motorcycle, Bicycle, E-Scooters

블랙박스 영상 분석을 통한 토픽모델링 기반의 자율주행차 Edge Case 시나리오 구축 : 보행자, 이륜차, 자전거, 전동킥보드를 중심으로

Taejun KIM1
,
Hyeyeon CHU1
,
Jeong Ah Jang2
,
Hyunjeong Yun3
,
Junghwa KIM4*
김 태준1
,
주 혜연1
,
장 정아2
,
윤 현정3
,
김 정화4*
1Master Course, Department of Urban & Transportation Engineering, Kyonggi University, Gyeonggi 16227, Korea
2Research Professor, TOD-based Sustainable City/Transportation Research Center, Ajou University, Gyeonggi 16499, Korea
3Principal Researcher, Autonomous Driving Intelligence Research Laboratory, Electronics and Telecommunications Research Institute(ETRI), Daejeon 34129, Korea
4Assistant Professor, Department of Urban & Transportation Engineering, Kyonggi University, Gyeonggi 16227, Korea
1경기대학교 도시·교통공학과 석사과정
2아주대학교 TOD기반 지속가능 도시교통연구센터 연구교수
3한국전자통신연구원 자율주행지능연구실 책임연구원
4경기대학교 도시·교통공학과 조교수
*Corresponding Author
31 October 2025. pp. 554-575
PDF XML Citation
Abstract

Ensuring that autonomous vehicles can safely operate in complex road environments where they interact with various road users-including pedestrians, motorcycles, bicycles, and e-scooters-remains a critical challenge for the commercialization of autonomous driving technology. In particular, Edge Case scenarios, which are characterized by their low occurrence probability but high risk, are essential for validating the safety and reliability of autonomous vehicles. However, existing research has primarily focused on safety evaluation scenarios involving interactions between autonomous and conventional vehicles, with limited consideration of diverse road users such as pedestrians, motorcycles, bicycles, and scooters in real-world traffic environments. This study aims to develop Edge Case scenarios that address potential accident risks faced by autonomous vehicles. To achieve this, 2,410 dashcam dataset were collected from the youtube channel ‘HanMoonchul TV’. The data were processed through text preprocessing, topic modelling using the Latent Dirichlet Allocation (LDA) technique to extract key accident risk factors. The identified risk factors were classified according to the Pegasus 5-layer framework. Based on this classification, various Edge Case scenarios involving interactions with pedestrians, motorcycles, bicycles, and e-scooter were constructed. The results demonstrate that the designed scenarios enable autonomous vehicles to operate reliably even in unpredictable accident situations. These scenarios serve as tools for validating the safety and performance of autonomous vehicles. By employing a data-driven approach, this study addresses proposes a methodology for developing scenarios to evaluate safety in complex traffic environments.

Keywords
autonomous vehicle
edge case
pegasus 5-layer
safety evaluation scenarios
topic modelling

자율주행차가 다양한 도로 사용자와 혼재된 복잡한 도로 환경에서 안전하게 운행할 수 있도록 하는 것은 자율주행 기술 상용화를 위한 핵심 과제로 대두되고 있다. 특히, 발생 가능성은 낮지만 위험성이 높은 Edge Case 상황은 자율주행차의 안전성과 신뢰성을 검증하는 것이 매우 중요하다. 그러나 현재까지의 연구는 자율주행차와 차량 간 사고 중심의 안전성 평가 시나리오 구축에 집중되어 있으며, 실제 도로에서 자율주행차가 상호 작용할 보행자, 이륜차, 자전거, 킥보드 등의 다양한 객체를 고려한 연구는 부족한 실정이다. 이에 본 연구는 자율주행차와 해당 객체 간의 사고 위험상황을 다룰 수 있는 Edge Case 시나리오를 구축하는 것을 목표로 한다. 이를 위해, 유튜브 채널 '한문철 TV'에서 수집한 2,410건의 블랙박스 교통사고 데이터셋을 기반으로 LDA 토픽모델링 기법을 사용하여 주요 사고 위험 요인을 도출하였다. 도출된 위험 요인은 Pegasus 5-layer 포맷에 따라 체계적으로 분류하고, 이를 바탕으로 자율주행차와 보행자, 이륜차, 자전거, 전동킥보드 간 다양한 사고 시나리오를 구축하였다. 분석 결과, 자율주행차가 예측하기 어려운 사고 상황에서도 대응성을 향상할 수 있는 시나리오를 구축했으며, 이러한 시나리오는 실도로에서 자율주행차의 안전성과 성능 검증을 위한 도구로 활용될 수 있을 것으로 기대된다. 또한, 본 연구는 데이터 기반 접근 방식을 통해 복잡한 교통 환경에서 자율주행차의 안전성 평가를 위한 체계적인 시나리오 구축 방법을 제시하였다.

키워드
자율주행차
엣지 케이스
페가수스 5-레이어
안전성 평가 시나리오
토픽모델링
References
1

Blei D. M. (2012), Probabilistic Topic Models, Communications of the ACM, 55(4), Association for Computing Machinery, 77-84.

10.1145/2133806.2133826
2

Baek S. R., Kim Y. W., Yoo T. H., Kim C. H. (2024), Development of Database-Based Pedestrian Accident Scenarios and Analysis of Accident Prevention Effects from V2X Application, Journal of Korean Society of Intelligent Transport Systems, 23(5), The Korean Society of Intelligent Transport Systems, 274-292.

10.12815/kits.2024.23.5.274
3

Cao Y., Xiao L., Dong H., Wang Y., Wu X., Li P., Qiu Y. (2019), Typical Pre-Crash Scenarios Reconstruction for Two-Wheelers and Passenger Vehicles and its Application in Parameter Optimization of AEB System based on NAIS Database, In 26th International Technical Conference on the Enhanced Safety of Vehicles(ESV), 19-0272.

4

Chung S. (2024), A Study on the Methodology for Developing High-Risk Scenarios for the Validation of Autonomous Vehicles, Transactions of the Korean Society of Automotive Engineers, 32(3), The Korean Society of Automotive Engineers, 301-308.

10.7467/KSAE.2024.32.3.301
5

Durak U., Gupta S., Akgün K. (2023), Critical Scenario Techniques for Automated Vehicles: Literature Review, ASIM Workshop STS/GMMS/EDU Proceedings, 87-94.

10.11128/arep.21.a2124
6

Guo A., Zhou Y., Tian H., Fang C., Sun Y., Sun W., Gao X., Luu A. T., Liu Y., Chen Z. (2024), SoVAR: Building Generalizable Scenarios from Accident Reports for Autonomous Driving Testing, In Proceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering (ASE ’24), 1-13.

10.1145/3691620.3695037
7

Jo J. S., Kwak J. H., Ku D. G., Yoo Y. T., Lee S. J. (2022), Estimation of Leisure Behavior Using Unstructured Data: Focusing on Seoul City Wall, J. Korean Soc. Transp., 41(1), Korean Society of Transportation, 135-147.

10.7470/jkst.2023.41.1.135
8

Kalisvaart S. (2021a), Introducing Edge Cases, Presented at the EU-CSA ARCADE Workshop on Edge Cases, European Commission.

9

Kalisvaart S., Svanberg E., Barnard Y., Innamaa S., Sanz L., Bou J., Koskinen S. (2021b), Proceedings of Workshop on Edge Cases: ARCADE D2.4 Joint CCAM Networking Report, European Commission, Horizon 2020 Programme, 2-27.

10

Kang J. H., Ko W. R., Kim Y. J., Yoon J. G., Yun I. S. (2024), Development of Autonomous Vehicle Evaluation Scenarios Based on Car-to-Bicycle, Car-to-Pedestrian, and Car-to-Animal Traffic Accidents, Journal of Korean Society of Intelligent Transport Systems, 23(5), The Korean Society of Intelligent Transport Systems, 322-337.

10.12815/kits.2024.23.5.322
11

Kang M. H., Lee W. S., Hwang K. Y., Yoon Y. (2022), Vision Transformer for Detecting Critical Situations and Extracting Functional Scenario for Automated Vehicle Safety Assessment, Sustainability, 14(15), MDPI, 9680.

10.3390/su14159680
12

Kibalama D., Tulpule P. J. (2022), AV/ADAS Safety-Critical Testing Scenario Generation from Vehicle Crash Data, SAE Technical Papers, 1-9.

10.4271/2022-01-0104
13

Lee H. J., Kang M. H., Song, J. I., Hwang K. Y. (2023), Development of Autonomous Vehicle Traffic Accident Scenarios in Urban Areas Based on Real-World Accident Data Using Association Rule Mining, J. Korean Soc. Transp., 41(3), Korean Society of Transportation, 375-393.

10.7470/jkst.2023.41.3.375
14

Li P., Zhu X., Ren Y., Tan Z., Hu W., Zhang Y., Xu C. (2024), Generalization of Cut-in Pre-Crash Scenarios for Autonomous Vehicles based on Accident Data, Scientific Reports, 14, 17664.

10.1038/s41598-024-68263-439085290PMC11291781
15

Liu Q., Wang X., Liu S., Yu C., Glaser Y. (2024), Analysis of Pre-Crash Scenarios and Contributing Factors for Autonomous Vehicle Crashes at Intersections, Accident Analysis and Prevention, 195, Elsevier, 1-12.

10.1016/j.aap.2023.107383
16

Luettel T., Himmelsbach M., Wuensche H. J. (2012), Autonomous Ground Vehicles: Concepts and a Path to the Future, Proceedings of the IEEE, 100(1), IEEE, 1831-1840.

10.1109/JPROC.2012.2189803
17

Masmoudi M., Ghazzai H., Frikha M., Massoud Y. (2019), Object Detection Learning Techniques for Autonomous Vehicle Applications, IEEE International Conference on Vehicular Electronics and Safety (ICVES), 1-6.

10.1109/ICVES.2019.8906299
18

MIRA (2018), K-City Scenario Definition Consultancy.

19

MOLIT (2024), https://www.molit.go.kr/USR/NEWS/m_71/dtl.jsp?lcmspage=1&id=95089899, 2024.12.15.

20

Moradloo N., Mahdinia I., Khattak A. J. (2024), Safety in Higher Level Automated Vehicles: Investigating Edge Cases in Crashes of Vehicles Equipped with Automated Driving Systems, Accident Analysis & Prevention, 203, Elsevier, 107607.

10.1016/j.aap.2024.107607
21

Najm W. G., Smith J. D., Yanagisawa M. (2007), Pre-Crash Scenario Typology for Crash Avoidance Research (No. DOT-VNTSC-NHTSA-06-02), United States, National Highway Traffic Safety Administration.

22

NHTSA (2022), Summary Report : Standing General Order on Crash Reporting for Level 2 Advanced Driver Assistance Systems, United State, National Highway Traffic Safety Administration.

23

Oh J. S. (2015), Identifying Research Opportunities in the Convergence of Transportation and ICT Using Text Mining Techniques, J. Korean Soc. Transp., 22(4), Korean Society of Transportation, 93-110.

10.34143/jtr.2015.22.4.93
24

Park S. M., Park S. H., Jeong H. R., Yun I. S., So J. H. (2021a), Scenario-Mining for Level 4 Automated Vehicle Safety Assessment from Real Accident Situations in Urban Areas Using a Natural Language Process, Sensors, 21(20), MDPI, 6929.

10.3390/s2120692934696142PMC8537130
25

Park S. M., Lee H. P., So J. H., Yun I. S. (2021b), Analysis for Autonomous Vehicle Collision Using Text Embedding, Journal of Korean Society of Intelligent Transport Systems, 20(1), The Korean Society of Intelligent Transport Systems, 160-173.

10.12815/kits.2021.20.1.160
26

Park S. M., So J. H., Ko H. G., Jeong H. R., Yun I. S. (2018), Development of Safety Evaluation for Automated Vehicle Tests Using 5-Layer Format (Case of the Community Road), Journal of Korean Society of Intelligent Transport Systems, 18(2), The Korean Society of Intelligent Transport Systems, 114-128.

10.12815/kits.2019.18.2.114
27

PEGASUS project, https://www.pegasusprojekt.de/en/home, 2024.12.17.

28

Rampf F., Grigoropoulos G., Malcolm P., Keler A., Bogenberger K. (2023), Modelling Autonomous Vehicle Interactions with Bicycles in Traffic Simulation, Frontiers in Future Transportation, 3, 894148.

10.3389/ffutr.2022.894148
29

Riedmaier S., Ponn T., Ludwig D., Schick B., Diermeyer F. (2020), Survey on Scenario-Based Safety Assessment of Automated Vehicles, IEEE Access, 8, IEEE, 87456-87458.

10.1109/ACCESS.2020.2993730
30

So J. H., Park I. S., Wee J. R., Park S. M., Yun I. S. (2019), Generating Traffic Safety Test Scenarios for Automated Vehicles using a Big Data Technique, KSCE Journal of Civil Engineering, 23(6), Korean Society of Civil Engineers, 2702-2712.

10.1007/s12205-019-1287-4
31

Sui B., Lubbe N., Bärgman J. (2019), A Clustering Approach to Developing Car-to-Two-Wheeler Test Scenarios for the Assessment of Automated Emergency Braking in China Using in-Depth Chinese Crash Data, Accident Analysis and Prevention, 132, Elsevier, 105242.

10.1016/j.aap.2019.07.018
32

Wang X., Peng Y., Xu T., Xu Q., Wu X., Xiang G., Yi S., Wang H. (2022), Autonomous Driving Testing Scenario Generation based on in-Depth Vehicle-to-Powered Two-Wheeler Crash Data in China, Accident Analysis & Prevention, 176, Elsevier, 106812.

10.1016/j.aap.2022.106812
33

Zhang Y., Ko S., Lee K. (2021), Autonomous Vehicle Scenario-based Safety Evaluation Using Improved Criticality Metrics, Accident Analysis & Prevention, 160, Elsevier, 106312.

Information
  • Publisher :Korean Society of Transportation
  • Publisher(Ko) :대한교통학회
  • Journal Title :Journal of Korean Society of Transportation
  • Journal Title(Ko) :대한교통학회지
  • Volume : 43
  • No :5
  • Pages :554-575
  • Received Date : 2025-01-20
  • Revised Date : 2025-02-26
  • Accepted Date : 2025-06-21
  • DOI :https://doi.org/10.7470/jkst.2025.43.5.554
Journal Informaiton Journal of Korean Society of Transportation Journal of Korean Society of Transportation
Archives
: 2015 Vol.34~
  • scopus
  • NRF
  • KOFST
  • crossref cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close