Ascertainment of On-Road Safety Errors Based on Video Review

Dawson, Jeffrey D; Ergun, UC; Anderson, Steven W; Dastrup, Elizabeth; Johnson, Amy M; Rizzo, Matthew · 2009 · OpenAlex-citations

DOI: 10.17077/drivingassessment.1352

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 objectively measuring on-road driving safety errors, particularly in elderly populations, where real-time assessment by a passenger is often impractical due to the need for instruction and safety monitoring. The authors propose a retrospective methodology based on video review of drives conducted in an instrumented vehicle. The study aims to validate a systematic taxonomy of safety errors and demonstrate its reliability and applicability for assessing driver performance in healthy elderly individuals, with potential extensions to patients with cognitive decline. The methodology utilized the ARGOS instrumented vehicle, which captured performance data (steering, pedal position, acceleration, speed) at 10 Hz and recorded video from four angles (driver, steering wheel, forward road, and center line) at 10 frames per second. The driving assessment involved a 35-mile route taking approximately 45 minutes, incorporating essential maneuvers and standardized cognitive challenges. A taxonomy of 15 general error categories and 76 specific error types, adapted from Iowa Department of Transportation standards, was employed. Professional driving instructors reviewed the video recordings to identify and code errors. Reliability was assessed through intra-rater and inter-rater correlations on a subset of tapes. The study sample consisted of 111 healthy elderly drivers (ages 65–89) who served as comparison subjects in broader studies of Alzheimer’s and Parkinson’s disease. The results indicated that the video review method was feasible and reliable, yielding a 95% intra-rater correlation and a 73% inter-rater correlation for total error counts. On average, drivers committed 35.8 safety errors per drive, with 2.1 classified as serious. Errors were observed in 13 of the 15 categories, with Lane Observance being the most frequent (mean 12.4 errors per drive), followed by Turns, Lane Changes, and Stop Signs. Of the 76 specific error types, 42 were observed, with touching the center line, failing to check blind spots during lane changes, and failing to stop completely at stop signs being the most common. The primary rater noted that the camera view of the center line and front left tire was most critical for error detection, suggesting improvements such as additional side-view cameras and color video. The significance of this work lies in establishing a standardized, archived method for assessing driving safety that correlates with neuropsychological tests of cognitive and motor skills. The authors conclude that this methodology is valid for predicting safety outcomes in patients with Alzheimer’s and Parkinson’s disease. It offers a robust tool for intervention studies and longitudinal monitoring of driving abilities in populations with declining cognitive function, providing a more objective alternative to subjective real-time ratings.

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.

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.

• human error taxonomy
• pedal misapplication
• pre crash contributing factors
• mci dementia driving
• naturalistic crash near crash
• 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