Automatic Safety Diagnosis in a Connected Vehicle Environment

Whalin, Robert W; Wu, Di; Tu, Shuang Z. · 2022 · ROSA P / Southeastern Transportation Research, Innovation, Development and Education Center (STRIDE)

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the critical need for real-time safety diagnosis in connected vehicle (CV) environments, motivated by the finding that human factors, specifically abnormal driving status, contribute to over 90% of traffic crashes. Traditional safety analysis methods, such as crash record statistics and simulation-based traffic conflict techniques, suffer from data latency, reporting inaccuracies, and an inability to provide immediate warnings. Furthermore, existing Advanced Driver Assistance Systems (ADAS) often rely on ego-vehicle sensors and focus on average driver behavior, failing to account for the unpredictable nature of abnormal drivers. The authors propose an Automatic Safety Diagnosis System in a Connected Vehicle Environment (ASDSCE) to identify near-crash events and generate warnings by leveraging Basic Safety Messages (BSMs). The ASDSCE is a computational pipeline built with Python on Visual Studio 2019, utilizing BSM data from CV pilot studies and evaluated against SHRP2 naturalistic driving study crash data. The system architecture comprises two main components: a cloud-based subsystem and an in-vehicle subsystem. The cloud component collects and stores historical BSMs to calculate individual vehicle thresholds for Key Performance Indicators (KPIs) in batch mode. The in-vehicle subsystem performs real-time analysis using two models: a Driving Anomaly Detection (DAD) model and a Conflict Identification Model (CIM). The DAD model operates in five modules: selecting KPIs, learning normal driving patterns, detecting outliers, determining abnormal events, and updating the system. The CIM identifies conflicts based on speed-distance profiles. A near-crash warning is triggered only when two conditions are met: a conflict is identified, and at least one involved driver is in an abnormal driving status. This dual-condition approach aims to reduce false alarms while ensuring warnings are issued with sufficient time for evasive action. The study found that the ASDSCE effectively identifies driving volatility and conflicts using solely BSM data, without relying on ego-vehicle sensors. By focusing on individual driver thresholds rather than aggregate levels, the system detects abnormal driving behaviors that precede crashes. The sensitivity analysis confirmed that the system parameters could be tuned to balance detection accuracy and response time. The use of BSMs allows for the extraction of essential safety thresholds while mitigating the storage burden of massive raw data volumes, as only relevant thresholds and short-term BSMs need to be retained. The significance of this work lies in its potential to enhance traffic safety by providing a supplementary collision warning tool that operates independently of ego-vehicle sensors. By integrating CV technology, traffic conflict analysis, and big data processing, the ASDSCE offers a proactive safety measure that can improve the safety of connected vehicles and support the market penetration of connected and autonomous vehicles. The system’s ability to detect abnormal drivers specifically addresses a major gap in current ADAS technologies. Future work includes pilot studies for further validation and upgrading the model to parallel processing to ensure reliable real-time performance.

Key finding

The ASDSCE successfully integrates individual-level driving anomaly detection with conflict identification to issue near-crash warnings based solely on BSM data.

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 bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

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.

• naturalistic crash near crash
• telematics crash prediction
• exposure measurement
• situational awareness
• incidence prevalence

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, validation psychometrics