Classification of Level 2 Driving Events Observed on Public Roads

Rao, Sughosh J.; Forkenbrock, Garrick J. · 2021 · ROSA P / United States. Department of Transportation. National Highway Traffic Safety Administration

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 report evaluates the performance of SAE Level 2 driving automation systems in real-world conditions, addressing the need to understand how these partial automation systems operate on public roads. The study focuses on three commercially available vehicles: a 2018 Cadillac CT6 (Super Cruise), a 2018 Nissan Leaf (ProPILOT Assist), and a 2018 Lexus LS500 (Lexus Safety System +A). The research was motivated by the variability in system implementations and the necessity for drivers to maintain full attention, requiring a detailed documentation of system behaviors and takeover requests. The experimental design involved driving the vehicles on three distinct routes: a 108-mile highway route, a 32.4-mile rural route, and a 63.1-mile mixed route comprising highway, rural, and residential roads. The Cadillac CT6 was tested only on the highway route due to its geofenced operational design domain, while the Nissan Leaf and Lexus LS500 were tested on all three routes. Professional drivers operated the vehicles with hands lightly touching or hovering above the steering wheel. Events were recorded using instrumented cameras and GPS, then classified into three categories: Type I (system-initiated termination and control transfer), Type II (noteworthy system behavior without control transfer, such as lane hugging or dithering), and Type III (driver-initiated override or unintended lane departure). The results revealed significant differences in system performance across vehicles and environments. The Cadillac CT6 averaged 22.7 Type I events per 100 miles on highways, often triggered by geofencing near ramps, with low rates of Type II (2.4) and Type III (0.7) events. The Nissan Leaf exhibited higher variability, with Type I events ranging from 10.5 to 23.7 per 100 miles across routes. Notably, the Nissan Leaf experienced frequent Type III events, averaging 13.0 to 27.5 per 100 miles, with nearly all rural and mixed route Type III events being lane departures. The Lexus LS500 showed the highest frequency of events, particularly Type III, which averaged 23.5 per 100 miles on highways but surged to 178.0 per 100 miles on rural routes, all of which were lane departures. Type II events for the Lexus also increased significantly in complex environments, reaching 32.6 per 100 miles on mixed routes. The study concludes by documenting the statistical occurrence of these events without drawing definitive conclusions regarding safety or consumer acceptance. It highlights that system availability and event frequency are heavily influenced by road type and system design, such as geofencing. The findings provide a baseline for understanding the operational limitations of current Level 2 systems, particularly the high incidence of lane departures and driver interventions in non-highway environments for non-geofenced systems. This data serves as a reference for future research into driver engagement and the reliability of automated driving features in diverse traffic scenarios.

Key finding

The study quantified the frequency of Type I, II, and III driving events for three Level 2 automation systems, revealing significant variations in event rates depending on the vehicle model and roadway type, with the Lexus LS500 exhibiting particularly high Type III event rates on rural and mixed routes.

Methodology

on_road

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
• urban rural setting
• situational awareness

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: conceptual framework