Identifying Scenarios in Field Data to Enable Validation of Highly Automated Driving Systems
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 validating highly automated driving systems, which requires identifying safety-critical scenarios in real-world traffic to inform software-in-the-loop simulations. Traditional distance-based validation methods are economically unfeasible, necessitating a shift toward scenario-based approaches that rely on realistic parameter distributions. The authors propose a method to identify predefined logical driving scenarios in field data by abstracting raw sensor measurements into qualitative models. This approach aims to reduce complex driving condition data to relevant variables, enabling the extraction of scenario probabilities and characteristics without the need for extensive training data required by machine learning methods. The methodology is built on knowledge-based metamodelling and pattern recognition. The authors define an ontology represented by a domain model using Unified Modeling Language classes, including actors, activities, events, and acts. A logical scenario is defined as a sequence of acts, where each act involves specific activities (e.g., lane changes, following) and events (e.g., threshold violations like distance or relative speed). The study focuses on the "cut-in" scenario as a case study, defining it through two primary events: "following" and "driving parallel." Field data is abstracted by classifying detected objects relative to the ego vehicle and identifying activities such as valid or invalid lane changes based on lateral position thresholds. The identification process involves chronologically comparing the abstracted field data against the logical scenario definitions to detect matches. The method was implemented in Python and validated using one-minute segments of highway test drive data accompanied by video recordings. The validation involved a test catalogue covering various events and activities, with results manually checked against video evidence to assess false positives and negatives. The results demonstrated that event identification worked correctly on straight road sections. However, the system struggled with curves, where vehicles overtaking on the outside were mistakenly identified as leading vehicles due to the sensor system’s reliance on lateral offset rather than road geometry. Additionally, the simplified implementation for detecting lane change durations introduced inaccuracies, sometimes identifying brief, invalid lane switches as valid. Despite these limitations, the core logic for identifying the cut-in scenario was confirmed to be functional. The significance of this work lies in providing a traceable, rule-based alternative to black-box machine learning approaches for scenario identification. By using a universal ontology, the method allows for the reduction of large volumes of measurement data to essential variables, facilitating the identification of various driving scenarios. The authors conclude that while sensor limitations currently hinder performance in curved sections, the proposed methodology is a suitable foundation for identifying logical scenarios in field data. Future work involves refining the abstraction of lane change activities and adapting the approach to handle more complex road geometries and a broader range of highway scenarios.
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 | 2026-08-09 |
| archive | success | openalex | — | — | 5 | 2026-08-09 |
| extract | success | cached | — | — | 3 | 2026-08-10 |
| clean | success | clean | — | — | 1 | 2026-08-09 |
| chunk | success | chunk | — | — | 1 | 2026-08-09 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-08-09 |
| enrich | success | semantic_scholar | — | — | 1 | 2026-08-09 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.6-27b-nvidia | summ-v5 | 2 | 2026-08-10 |
| tag | success | vector_similarity | — | — | 10 | 2026-08-11 |
| verify | success | — | — | — | 2 | 2026-08-10 |
Summary generated by qwen3.6-27b-nvidia on 2026-08-10; 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).
- Methodological Resource: dataset resource, tool software
- Theoretical Contribution: computational model