When should ADS Fallback? A Critique of the Solution for Improving the SAE Conditional Driving Automation Definition

Liang, Yaowei; Ayas, Suzan; Jamieson, Greg A. · 2025 · Crossref

DOI: 10.21203/rs.3.rs-7687979/v1

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 critiques the proposal by Inagaki and Sheridan (2019) to add an intermediate level between SAE Level 3 and Level 4 driving automation. Inagaki and Sheridan argued that the existing SAE J3016 taxonomy was incomplete because it did not adequately address scenarios where drivers fail to respond to Requests to Intervene (RTI). They proposed a new level where the Automated Driving System (ADS) waits several seconds for driver input before executing a fallback maneuver. The authors of this study challenge that solution by demonstrating that waiting for driver intervention increases risk as the vehicle continues to approach the Operational Design Domain (ODD) boundary. The authors extend Inagaki and Sheridan’s mathematical framework by modeling available intervention time (AIT) as a continuous, time-dependent variable. Unlike the previous model, which treated takeover benefits and costs as fixed values, this study incorporates the increasing risk cost associated with the vehicle traveling closer to hazards during the AIT. The expected utility function accounts for the probability of successful driver intervention, the quality of that intervention, the constant benefit of ADS fallback, and the strictly increasing risk cost of delay. The analysis considers scenarios where the ADS has a fallback function and where drivers can veto the RTI. The results show that the expected utility function initially decreases and then increases as AIT grows, creating a local minimum. This indicates that shortly after an RTI is issued, the risk of continuing to approach the hazard outweighs the potential benefit of waiting for a driver response. Consequently, for an ADS equipped with a fallback function, executing the fallback immediately rather than waiting for driver input maximizes expected utility. The study finds that delaying fallback lowers utility, particularly in emergency situations where the time to reach the ODD boundary is short. These findings challenge the necessity of adding an intermediate level between SAE Level 3 and Level 4. The authors conclude that if an ADS has fallback capability, it should execute that function as early as possible to minimize risk, aligning more closely with the definition of SAE Level 4. The presence or absence of ADS fallback capability, rather than the timing of its execution relative to driver response, should be the primary criterion for defining automation levels. This approach simplifies the taxonomy and prioritizes safety by avoiding the risks associated with prolonged exposure to hazards during driver takeover attempts.

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-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• driverless ads
• takeover transitions
• automation
• automation surprise
• teleoperation remote driving

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).

• Theoretical Contribution: conceptual framework, computational model