Abnormal Driving Detection Using GPS Data
DOI: 10.1109/honet59747.2023.10374718
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 Abnormal Driving Detection (ADD) using real-world GPS data, aiming to distinguish between normal and abnormal driving behaviors. The research is motivated by the need for improved driver safety, accurate insurance risk assessment, personalized interventions, and enhanced emergency response capabilities. Unlike prior studies that often rely on synthetic data or generative models like Hidden Markov Models, this work utilizes authentic GPS records captured at one-second intervals from actual vehicles, providing a more accurate representation of real-world driving patterns. The methodology employs an integrated approach consisting of data preprocessing, dimensionality reduction, and clustering. The input data includes Speed Over Ground (SOG), Course Over Ground (COG), longitude, and latitude. These features are aggregated into minute-level segments to allow for fine-grained analysis. The authors calculate specific metrics such as Speed Change (indicating hard braking or acceleration), COG Change (indicating oversteering or lane changes), Vehicle Direction, and Turn Sharpness. To handle the high dimensionality of the feature space, Singular Value Decomposition (SVD) is applied to reduce the data matrix to two principal latent features. Subsequently, the K-means clustering algorithm is used to partition the dataset into two distinct homogeneous groups, classifying driving instances as either normal or abnormal based on their behavioral patterns. The study demonstrates the effectiveness of this pipeline through a toy example and experimental results on real-world datasets, including data from elderly drivers. The SVD process successfully reduces the complexity of the driving records, revealing similarities and distinct patterns among different driving segments. The K-means clustering then clearly separates these patterns into two clusters, validating the method's ability to identify anomalies such as erratic speed changes, aggressive maneuvers, or weaving. The results confirm that the proposed approach can accurately discern abnormal driving behaviors from normal ones using only GPS-derived features. The significance of this work lies in its practical applications across multiple domains. In transportation safety, the system can serve as a vigilant watchdog to identify hazardous situations and provide corrective interventions. For insurance companies, it offers a data-driven method to assess risk and set premiums more accurately. Additionally, the technology supports elderly care by monitoring driving behavior for individuals with dementia, alerting caregivers to deviations from standard patterns. The use of authentic GPS data and the novel combination of SVD and K-means clustering advance the field of ADD, offering a scalable and practical solution for real-time monitoring and analysis of driver behavior.
Key finding
The proposed methodology using Singular Value Decomposition and K-means clustering on GPS data effectively distinguishes between normal and abnormal driving behaviors.
Methodology
dataset
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. Discovered via author_sweep_intake on 2026-05-27.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-27 |
| archive | success | canonical_url | — | — | 11 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | semantic_scholar | — | — | 2 | 2026-06-04 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; 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