All Issue

2026 Vol.44, Issue 5 Preview Page

Article

Scenario Classification for Vehicle Trajectory Prediction in Urban Environments using Multi-Agent and Geometry Context

도심지 주행환경의 이동 궤적 예측을 위한 다중 객체 및 기하 구조 기반 시나리오 분류

Beomjin PARK1
,
Simon OH2
,
Jinwon YOON3*
박 범진1
,
오 시몬2
,
윤 진원3*
1Graduate Student, Department of Computer Science and Software Engineering, Korea University, Sejong 30019, Korea
2Associate Professor, Department of Mobility Science & Engineering, Korea University, Sejong 30019, Korea
3Assistant Professor, Department of Mobility Science & Engineering, Korea University, Sejong 30019, Korea
1고려대학교 컴퓨터정보학과 석박사통합과정
2고려대학교 미래모빌리티학과 부교수
3고려대학교 미래모빌리티학과 조교수
*Corresponding Author
31 October 2026. pp. 973-987
PDF XML Citation
Abstract

Recent prediction technology for multi-agent movement trajectories in autonomous driving has greatly advanced along with the development of large-scale datasets and learning-based models. However, the application of these prediction models to real roads can be limited, and consistently maintaining high prediction performance across various prediction scenarios is still pointed out as a major limitation. This can be interpreted as originating from the training environment of existing prediction models for biased scenarios. Therefore, this study hierarchically classifies scenario-level factors into two factors, surrounding agents and road geometry, and seeks to identify the limitations and solutions of prediction models based on open datasets such as Argoverse by comparatively analyzing prediction performance by scenario type. A conditional variational autoencoder (CVAE) architecture was used and selected as the prediction model, and prediction accuracy was evaluated by measuring the average and final position errors (minADE, minFDE) with respect to the prediction target. The experimental results showed that the agent factor mainly had a large effect on overall prediction performance, whereas the intersection existence caused an increase in performance variance. In particular, in scenarios including intersections and high lane complexity/road connectivity, performance variability greatly increased, suggesting that prediction performance for multi-agent movement trajectories under complex geometric conditions may appear low despite high overall performance values. Accordingly, to improve the stability of movement trajectory prediction in urban driving environments, measures such as scenario dataset augmentation for complex geometric structures are required, and it is necessary to prepare prediction strategies that can embed situational/environmental context in the prediction process. The scenario classification framework proposed in this study can be used in comparative studies between prediction models, and it is expected to contribute to performance improvement through follow-up measures such as model calibration when preparing new prediction solutions.

Keywords
autonomous driving
autonomous vehicle dataset
Conditional Variational Autoencoder (CVAE)
scenario classification
trajectory prediction

최근 자율주행에서의 다중 객체 이동 궤적에 대한 예측 기술은 대규모 데이터셋과 학습 기반 모델의 발전과 함께 크게 발전하였다. 하지만 이들 예측 모델의 실도로 적용은 제한될 수 있으며 높은 예측 성능을 다양한 예측 시나리오에 대해 일관성 있게 유지하는 부분은 여전히 큰 한계점으로 지적된다. 이는 기존 예측 모델들의 편중된 시나리오에 대한 학습 환경에서 기인한 것으로 해석될 수 있다. 이에 본 연구에서는 시나리오 수준의 요인을 두 가지 요인(주변 객체 및 도로 기하구조)으로 계층적으로 분류하고 예측 성능을 시나리오 타입별로 비교･분석 함으로써 Argoverse와 같은 오픈 데이터셋 기반의 예측 모델의 한계점과 해결방안을 모색하고자 한다. 조건부 변분 오토인코더(CVAE) 아키텍처를 활용하여 예측 모델로 선정되었으며 예측 대상과의 평균 및 최종 위치 오차(minADE, minFDE)를 측정함으로써 예측 정확도를 평가한다. 실험 결과, 객체 인자는 주로 예측 성능의 전반에 큰 영향을 주는 반면, 기하구조 요인은 성능 분산 증가를 야기하는 것으로 나타났다. 특히 교차로를 포함하지 않는 시나리오에서는 minADE@6와 minFDE@6는 각각 1.18±1.15m와 3.10±3.41m였으나, 교차로 포함 시나리오에서는 각각 1.37±1.39m와 3.66±4.05m로 증가하였다. 이는 교차로와 같은 복잡한 기하 조건에서는 다중 객체의 이동 궤적 예측 오차와 성능 변동성이 증가할 수 있음을 시사한다. 이에 도시 주행 환경에서의 이동궤적 예측의 안정성을 향상시키기 위해서는 복잡한 기하구조에 대한 시나리오 데이터셋 증강 등의 조치가 요구되며 예측 과정에서 상황/환경적 맥락을 임베딩할 수 있는 예측 전략을 마련할 필요가 있다. 본 연구에서 제시된 시나리오 분류체계는 예측 모델 간 비교 연구에서 활용될 수 있으며 새로운 예측 솔루션 마련 시 모델 보정 등 후속적인 조치를 통한 성능 향상에 기여할 수 있을 것으로 기대된다.

키워드
자율주행
자율주행 데이터셋
CVAE
시나리오 분류
이동 궤적 예측
References
1

Alahi A., Goel K., Ramanathan V., Robicquet A., Fei-Fei L., Savarese S. (2016), Social LSTM: Human Trajectory Prediction in Crowded Spaces, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 961-971.

10.1109/CVPR.2016.110
2

Caesar H., Bankiti V., Lang A. H., Vora S., Liong V. E., Xu Q. et al. (2020), nuScenes: A Multimodal Dataset for Autonomous Driving, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 11621-11631.

10.1109/CVPR42600.2020.01164
3

Chang M. F., Lambert J., Sangkloy P., Singh J., Bak S., Hartnett A. et al. (2019), Argoverse: 3D Tracking and Forecasting with Rich Maps, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 8748-8757.

10.1109/CVPR.2019.00895
4

Chen C., Pourkeshavarz M., Rasouli A. (2024), CRITERIA: A New Benchmarking Paradigm for Evaluating Trajectory Prediction Models for Autonomous Driving, Proceedings of the 2024 IEEE International Conference on Robotics and Automation (ICRA), IEEE, 8265-8271.

10.1109/ICRA57147.2024.10610911
5

Choi S., Kim J., Yu H., Ka D., Yeo H. (2019), Deep-learning Based Urban Vehicle Trajectory Prediction, J. Korean Soc. Transp., 37(5), Korean Society of Transportation, 422-429.

10.7470/jkst.2019.37.5.422
6

Codevilla F., Santana E., López A. M., Gaidon A. (2019), Exploring the Limitations of Behavior Cloning for Autonomous Driving, Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, 9329-9338.

10.1109/ICCV.2019.00942
7

Deo N., Trivedi M. M. (2018), Multi-Modal Trajectory Prediction of Surrounding Vehicles with Maneuver Based LSTMs, Proceedings of the IEEE Intelligent Vehicles Symposium (IV), IEEE, 1179-1184.

10.1109/IVS.2018.8500493
8

Department for Transport (2022), Connected and Automated Mobility 2025: Realising the Benefits of Self-driving Vehicles in the UK, Policy Paper, UK Government.

9

Ettinger S., Cheng S., Caine B., Liu C., Zhao H., Pradhan S. et al. (2021), Large Scale Interactive Motion Forecasting for Autonomous Driving: The Waymo Open Motion Dataset, Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, 9710-9719.

10.1109/ICCV48922.2021.00957
10

Farid A., Veer S., Ivanovic B., Leung K., Pavone M. (2023), Task-Relevant Failure Detection for Trajectory Predictors in Autonomous Vehicles, Proceedings of the 6th Conference on Robot Learning (CoRL), Proceedings of Machine Learning Research, 205, 1959-1969.

11

Gao J., Sun C., Zhao H., Shen Y., Anguelov D., Li C. et al. (2020), VectorNet: Encoding HD Maps and Agent Dynamics from Vectorized Representation, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 11525-11533.

10.1109/CVPR42600.2020.01154
12

Grand View Research (2024), Autonomous Vehicles Market Size, Share & Trends Analysis Report by Application, by Region, and Segment Forecasts 2024-2030, Industry Report, Grand View Research.

13

Gu J., Sun C., Zhao H. (2021), DenseTNT: End-to-End Trajectory Prediction from Dense Goal Sets, Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, 15303-15312.

10.1109/ICCV48922.2021.01502
14

Hochreiter S., Schmidhuber J. (1997), Long Short-Term Memory, Neural Computation, 9(8), MIT Press, 1735-1780.

10.1162/neco.1997.9.8.1735
15

Houenou A., Bonnifait P., Cherfaoui V., Yao W. (2013), Vehicle Trajectory Prediction Based on Motion Model and Maneuver Recognition, Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, 4363-4369.

10.1109/IROS.2013.6696982
16

Kim T. J., Oh C., Kang K. P. (2011), Prediction of Rear-end Crash Potential Using Vehicle Trajectory Data, J. Korean Soc. Transp., 29(3), Korean Society of Transportation, 73-82.

17

Lee J. K., Yang I. C., Kim H. S. (2025), A Study on Efficient Methodology of Using Transformers to Predict Vehicle Turning Trajectories at Intersections, Journal of Korean Society of Intelligent Transport Systems, Korean Society of Intelligent Transport Systems, 24(1), 26-37.

10.12815/kits.2025.24.1.26
18

Lee N., Choi W., Vernaza P., Choy C. B., Torr P. H. S., Chandraker M. (2017), DESIRE: Distant Future Prediction in Dynamic Scenes With Interacting Agents, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 336-345.

10.1109/CVPR.2017.233
19

Li X., Ying X., Chuah M. C. (2019), GRIP: Graph-based Interaction-aware Trajectory Prediction, Proceedings of the 2019 IEEE Intelligent Transportation Systems Conference (ITSC), IEEE, 3960-3966.

10.1109/ITSC.2019.8917228
20

Liang M., Yang B., Hu R., Chen Y., Liao R., Feng S. et al. (2020), Learning Lane Graph Representations for Motion Forecasting, Proceedings of the European Conference on Computer Vision (ECCV), Springer, 541-556.

10.1007/978-3-030-58536-5_32
21

National Highway Traffic Safety Administration (2017), Automated Driving Systems 2.0: A Vision for Safety, Report No.DOT HS 812 442, U.S. Department of Transportation.

22

Nayakanti N., Al-Rfou R., Zhou A., Goel K., Refaat K. S., Sapp B. (2023), Wayformer: Motion Forecasting via Simple & Efficient Attention Networks, Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), IEEE, 2980-2987.

10.1109/ICRA48891.2023.10160609
23

Ngiam J., Caine B., Vasudevan V., Zhang Z., Chiang H. T. L., Ling J. et al. (2022), Scene Transformer: A Unified Architecture for Predicting Multiple Agent Trajectories, Proceedings of the International Conference on Learning Representations (ICLR).

24

Salzmann T., Ivanovic B., Chakravarty P., Pavone M. (2020), Trajectron++: Dynamically-Feasible Trajectory Forecasting With Heterogeneous Data, Proceedings of the European Conference on Computer Vision (ECCV), Springer, 683-700.

10.1007/978-3-030-58523-5_40
25

Schwarting W., Alonso-Mora J., Rus D. (2018), Planning and Decision-Making for Autonomous Vehicles, Annual Review of Control, Robotics, and Autonomous Systems, 1(1), Annual Reviews, 187-210.

10.1146/annurev-control-060117-105157
26

Shi S., Jiang L., Dai D., Schiele B. (2022), Motion Transformer with Global Intention Localization and Local Movement Refinement, Advances in Neural Information Processing Systems (NeurIPS), 35, 6531-6543.

10.52202/068431-0473
27

Stoler B., Navarro I., Jana M., Hwang S., Francis J., Oh J. (2024), SafeShift: Safety-Informed Distribution Shifts for Robust Trajectory Prediction in Autonomous Driving, Proceedings of the 2024 IEEE Intelligent Vehicles Symposium (IV), IEEE, 1179-1186.

10.1109/IV55156.2024.10588828
28

Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A. N. et al. (2017), Attention Is All You Need, Advances in Neural Information Processing Systems (NeurIPS), Curran Associates, Inc., 5998-6008.

29

Wilson B., Qi W., Agarwal T., Lambert J., Singh J., Khandelwal S. et al. (2021), Argoverse 2: Next Generation Datasets for Self-driving Perception and Forecasting, Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks (NeurIPS Datasets and Benchmarks), NeurIPS.

30

Xu C., Ding W., Lyu W., Liu Z., Wang S., He Y. et al. (2022), SafeBench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehicles, Advances in Neural Information Processing Systems (NeurIPS), 35, Datasets and Benchmarks Track., 25667-25682.

10.52202/068431-1861
31

Yan Z., Li P., Fu Z., Xu S., Shi Y., Chen X. et al. (2023), INT2: Interactive Trajectory Prediction at Intersections, Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, 8536-8547.

10.1109/ICCV51070.2023.00784
32

Zhao H., Gao J., Lan T., Sun C., Sapp B., Varadarajan B. et al. (2021), TNT: Target-driven Trajectory Prediction, Proceedings of the 2020 Conference on Robot Learning (CoRL), Proceedings of Machine Learning Research, 155, 895-904.

33

Zhou Z., Wang J., Li Y. H., Huang Y. K. (2023), Query-Centric Trajectory Prediction, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 17863-17873.

10.1109/CVPR52729.2023.01713
34

Zhou Z., Ye L., Wang J., Wu K., Lu K. (2022), HiVT: Hierarchical Vector Transformer for Multi-Agent Motion Prediction, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 8823-8833.

10.1109/CVPR52688.2022.00862
Information
  • Publisher :Korean Society of Transportation
  • Publisher(Ko) :대한교통학회
  • Journal Title :Journal of Korean Society of Transportation
  • Journal Title(Ko) :대한교통학회지
  • Volume : 44
  • No :5
  • Pages :973-987
  • Received Date : 2026-07-13
  • Revised Date : 2026-07-17
  • Accepted Date : 2026-08-17
  • DOI :https://doi.org/10.7470/jkst.2026.44.5.973
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