Evaluation of car-pedestrian incident data conversion technique in near-miss incident database to use driving simulator scenarios
DOI: 10.1299/jsmetld.2016.25.1305
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the challenge of reducing car-pedestrian accidents by developing a method to convert real-world near-miss incident data into driving simulator scenarios. Although traffic fatalities in Japan have decreased, the total number of accidents and injuries remains high, with pedestrian incidents constituting the largest share of fatalities. To improve driver behavior evaluation and support system development, the authors sought to bridge the gap between realistic accident analysis and controlled experimental environments. Real-world testing is hazardous and difficult to prepare, while traditional simulator studies often lack realism due to extreme or artificial scenarios. The study aims to utilize a near-miss incident database to create simulator scenarios that accurately reflect real-world conditions, allowing for detailed analysis of pedestrian crossing behaviors and driver responses. The methodology involves extracting data from the Tokyo University of Agriculture and Technology Near-Miss Database, which contains approximately 110,000 incidents recorded via dashcams on taxis. The researchers focused on 50 specific cases involving pedestrians: 5 actual accidents and 45 near-miss incidents where pedestrians darted out or crossed roads. They analyzed these cases to identify key factors such as pedestrian emergence positions, vehicle speeds, and environmental obstructions. Two representative patterns were selected for simulation: Pattern A, where a pedestrian crosses from behind a stationary vehicle on a wide road, and Pattern B, where a pedestrian emerges from behind a building on a narrow road. To reconstruct these scenes, the team integrated GPS data, vehicle speed, acceleration, and forward-facing video footage with map information from the Geospatial Information Authority of Japan. They calculated positions and orientations for the ego vehicle and other traffic participants at one-second intervals, using interpolation for obscured movements. These data were implemented into a single-seat driving simulator equipped with an HMD, steering, accelerator, and brake inputs, as well as audio and vibration feedback. The results demonstrate that the near-miss database can be effectively converted into simulator scenarios with high fidelity. Visual comparisons between the original dashcam footage and the simulator’s CG output showed general consistency in the positioning of pedestrians and opposing vehicles, with minor discrepancies attributed to differences in camera field of view and height. Furthermore, the simulator allowed for the manipulation of scenario variables; by delaying the pedestrian’s emergence by approximately 0.53 seconds, the researchers successfully created a more dangerous situation that altered the spatial relationship between the vehicle and pedestrian, potentially requiring evasive steering. This confirmed that the system can reproduce real-world incidents and modify timing to assess varying levels of collision risk. The significance of this work lies in providing a practical framework for creating realistic driving simulator scenarios from real-world near-miss data. By enabling the reproduction of actual incident environments and the adjustment of critical variables like timing and position, this method facilitates the evaluation of driver behavior and the effectiveness of safety support systems under realistic conditions. The authors conclude that this approach supports future research in hazard perception training and the analysis of specific accident types, offering a more grounded alternative to artificial simulator experiments.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-10 |
| archive | success | canonical_url | — | — | 1 | 2026-06-25 |
| extract | success | cached | — | — | 2 | 2026-06-25 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| enrich | success | openalex | — | — | 1 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-10 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-25 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: crash risk outcomes
- Methodological Resource: dataset resource, tool software