Driver Behaviour Profiling Using Dynamic Bayesian Network
DOI: 10.5815/ijmecs.2018.07.05
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 monitoring and analyzing vehicle driver behavior to enhance road safety. As vehicle numbers and road network complexity increase, understanding driving styles—characterized by acceleration, braking, and cornering patterns—becomes critical. The authors propose a cost-effective, easy-to-implement model that profiles drivers using only Global Positioning System (GPS) data, avoiding the need for complex inertial sensors or extensive calibration required by smartphone-based solutions. The study is motivated by the hypothesis that probabilistic methodologies, specifically Dynamic Bayesian Networks (DBNs), are suitable for determining driving styles and operational environments, potentially encouraging safer driving through feedback. The proposed system comprises four main components: an instrumented vehicle with a GPS on-board unit, a central server, a driver with a mobile device, and network services (GPRS/GSM). The vehicle collects real-time GPS data, specifically speed, altitude, direction, and signal strength, which is transmitted to a central server. The server processes this data using a 2-Time-slice Bayesian Network (2TBN), a probabilistic graphical model that maps variables and their conditional dependencies over time. The model classifies driver profiles into two sets: behavioral attributes (normal vs. harsh acceleration, braking, and cornering) and environmental conditions (meander, straight, up-hill, down-hill). Profiling metrics are derived from GPS parameters; for instance, acceleration and braking are deemed "harsh" if the distance covered exceeds 5 meters per square second, based on stopping sight distance calculations. The system generates a profile summary and sends notifications to drivers via SMS or email at the end of a journey or day. The study validates the model through face validity with industry experts and by leveraging pilot study results from previous work by the authors. The 2TBN model defines 64 possible states per time-slice, calculating probabilities for behavior and environment based on changes in speed, altitude, and direction relative to GPS signal strength. The results indicate that the 2-Time-slice Bayesian Network is suitable for generating driver profiles using only the four specified GPS parameters. The model successfully maps the dynamic nature of GPS data as a time series, allowing for the calculation of behavioral and environmental probabilities for each time-slice. The authors note that while the current state space is manageable, adding variables like obstacles would significantly increase complexity. The significance of this work lies in providing an affordable, scalable solution for driver profiling that supports Intelligent Transportation Systems. Unlike previous models that rely on expensive hardware or complex smartphone calibration, this approach uses standard GPS data, making it accessible for applications in vehicle driver recruiting, insurance companies, and transport authorities. By providing drivers with feedback on their behavior and operational environment, the model aims to reduce risky driving behaviors. The study concludes that probabilistic graphical models offer a robust alternative to fuzzy logic and fixed-threshold methods for driver behavior analysis, particularly in contexts where low-cost infrastructure is preferred.
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 | OpenAlex-citations | — | — | 1 | 2026-06-19 |
| archive | success | unpaywall | — | — | 2 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| 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 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; 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: observational prevalence
- Methodological Resource: dataset resource
- Theoretical Contribution: computational model