arxiv: 2605.06439 · v1 · submitted 2026-05-07 · 💻 cs.CY · cs.CV· cs.ET

Recognition: unknown

From Review to Design: Ethical Multimodal Driver Monitoring Systems for Risk Mitigation, Incident Response, and Accountability in Automated Vehicles

Bilal Khana , Waseem Shariff , Rory Coyne , Muhammad Ali Farooq , Peter Corcoran

Authors on Pith no claims yet

Pith reviewed 2026-05-08 04:47 UTC · model grok-4.3

classification 💻 cs.CY cs.CVcs.ET

keywords driver monitoring systemsethical frameworkautomated vehiclesprivacyconsentfairnesstransparencyaccountability

0 comments

The pith

The paper proposes a modular ethical design framework that converts broad regulatory principles into specific, actionable features for driver monitoring systems in automated vehicles.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper examines the ethical and legal issues arising from AI-based driver monitoring systems that use cameras and sensors to track attention and readiness in self-driving cars. General regulations do not provide enough detail for the privacy, fairness, and consent problems unique to in-cabin monitoring. The authors review these shortcomings and introduce a modular framework with built-in consent options, bias-reduced model training, explanation tools, and mental health protections for drivers. They also describe ways to analyze risks and respond to failures while maintaining accountability. Readers should care because these systems will soon be required for safe automation, and poor design could lead to widespread distrust or legal issues.

Core claim

The paper's core contribution is a modular ethical design framework for multimodal Driver Monitoring Systems. This framework takes high-level principles from existing regulations and turns them into practical design elements, such as user-configurable consent mechanisms, fairness-aware model development, transparency and explainability tools, and safeguards for driver emotional well-being. It further provides a risk analysis and failure mitigation strategy focused on incident response and accountability in the DMS context.

What carries the argument

The modular ethical design framework, which structures ethical requirements into separate, implementable modules for consent management, model fairness, system transparency, emotional safeguards, and risk response in driver monitoring.

Load-bearing premise

That identified gaps in regulations can be addressed adequately by applying a modular design framework without the need for new legislation or conflicts with current rules.

What would settle it

Deploying a DMS prototype built according to the framework and submitting it for review under GDPR or the EU AI Act to see if it resolves the identified applicability gaps without additional regulatory changes.

Figures

Figures reproduced from arXiv: 2605.06439 by Bilal Khana, Muhammad Ali Farooq, Peter Corcoran, Rory Coyne, Waseem Shariff.

**Figure 1.** Figure 1: Categorisation of Ethical Framework based on different interfaces, e.g. AVs and view at source ↗

**Figure 2.** Figure 2: Scope of this paper shown in figure1, and organizing ethical challenges at the relevant interface levels: (i) between AVs and DMS, (ii) between DMS and the human driver, and (iii) between AVs and the human driver view at source ↗

**Figure 3.** Figure 3: Proposed ethical framework for DMS showing GDPR-compliant data steward view at source ↗

**Figure 4.** Figure 4: Overview of how the proposed framework addresses key ethical challenges in view at source ↗

read the original abstract

As vehicles transition toward higher levels of automation, Driver Monitoring Systems (DMS) have become essential for ensuring human oversight, safety, and regulatory compliance in a vehicle. These systems rely on multimodal sensing and AI-driven inference to assess driver attention, cognitive state, and readiness to take control. While technologically promising, their deployment introduces a complex set of ethical and legal challenges - ranging from privacy and consent to data ownership and algorithmic fairness. While overarching frameworks such as the GDPR, EU AI Act, and IEEE standards offer important guidance, they lack the specificity required for addressing the unique risks posed by in-cabin sensing technologies. This paper adopts a review-to-design perspective, critically examining existing regulatory instruments and ethical frameworks -- such as the GDPR, the EU AI Act, and IEEE guidelines -- and identifying gaps in their applicability to the distinctive risks posed by multimodal, AI-enabled in-cabin monitoring. Building on this review, we propose a modular ethical design framework tailored specifically to Driver Monitoring Systems. The framework translates high-level principles into actionable design and deployment guidance, including user-configurable consent mechanisms, fairness-aware model development, transparency and explainability tools, and safeguards for driver emotional well-being. Finally, the paper outlines a risk analysis and failure mitigation strategy, emphasizing proactive incident response and accountability mechanisms tailored to the DMS context. Together, these contributions aim to inform the development of transparent, trustworthy, and human-centered driver monitoring systems for next-generation autonomous vehicles.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper synthesizes regulatory gaps for in-cabin DMS into a modular ethical framework but stays conceptual without validation or concrete compliance checks.

read the letter

This paper reviews GDPR, the EU AI Act, and IEEE guidelines, notes their lack of specificity for multimodal driver monitoring in automated vehicles, and proposes a modular ethical design framework with components for consent, fairness-aware models, explainability, emotional safeguards, plus risk analysis and accountability steps. The modular structure is a reasonable way to break down the issues into design choices rather than leaving everything at the principle level. The synthesis of the regulatory landscape in one place is useful for anyone trying to connect broad rules to in-cabin sensing risks. The addition of emotional well-being and incident response elements shows some attention to practical deployment concerns that often get ignored. The main limitation is that the work remains a high-level proposal. It identifies gaps but provides no detailed mappings to specific regulatory articles, no case studies showing how the modules would resolve conflicts such as safety overrides versus consent withdrawal, and no evaluation of whether the framework actually satisfies or avoids clashing with rules like GDPR Article 9 or the AI Act's high-risk biometric obligations. Without those steps the claim that design choices alone can close the gaps stays unverified. This is the sort of paper that would interest AV engineers, ethicists, and regulators who need a structured starting point for thinking through DMS ethics. A reader focused on applied AI governance in transportation could pull some value from the breakdown of components. It deserves peer review because the topic is current and the structure is clear, though reviewers would probably push for more concrete examples and regulatory interaction analysis to make the contribution stronger.

Referee Report

2 major / 2 minor

Summary. The paper reviews ethical and legal challenges in multimodal Driver Monitoring Systems (DMS) for automated vehicles, identifies gaps in the specificity of GDPR, EU AI Act, and IEEE guidelines for in-cabin sensing risks, and proposes a modular ethical design framework that translates high-level principles into components including user-configurable consent mechanisms, fairness-aware model development, transparency/explainability tools, and emotional well-being safeguards. It concludes with a risk analysis, failure mitigation strategy, and accountability mechanisms for incident response.

Significance. If the framework's components can be shown to resolve the identified regulatory gaps without introducing conflicts or safety trade-offs, the work would offer constructive, actionable guidance for ethical DMS design in autonomous vehicles, helping bridge abstract standards with practical deployment needs in AI ethics and transportation systems.

major comments (2)

[Proposed modular ethical design framework] In the section proposing the modular ethical design framework (as summarized in the abstract), the claim that components such as user-configurable consent and fairness-aware models adequately address gaps in GDPR and the EU AI Act is not supported by concrete mappings or exception analyses. For instance, no discussion is provided on compatibility with GDPR Article 9 special-category data rules for biometric processing or EU AI Act high-risk obligations for in-cabin monitoring systems, nor on how consent withdrawal would be overridden in safety-critical scenarios without impairing vehicle control.
[Risk analysis and failure mitigation strategy] The risk analysis and failure mitigation strategy (outlined in the final contribution) presents proactive incident response and accountability mechanisms at a high level but does not include case-specific mappings, compliance checklists, or evaluation criteria to verify that the safeguards prevent the very regulatory conflicts identified earlier in the review.

minor comments (2)

[Abstract] The abstract would benefit from a brief outline of the paper's section structure to help readers navigate the transition from review to design proposal.
[Introduction / Background] Terminology such as 'multimodal sensing' and 'in-cabin monitoring' could be defined more precisely on first use to ensure consistency for readers outside the immediate subfield.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed report. We address each major comment below and commit to revisions that will enhance the specificity and actionability of the proposed framework without altering the manuscript's core contributions.

read point-by-point responses

Referee: [Proposed modular ethical design framework] In the section proposing the modular ethical design framework (as summarized in the abstract), the claim that components such as user-configurable consent and fairness-aware models adequately address gaps in GDPR and the EU AI Act is not supported by concrete mappings or exception analyses. For instance, no discussion is provided on compatibility with GDPR Article 9 special-category data rules for biometric processing or EU AI Act high-risk obligations for in-cabin monitoring systems, nor on how consent withdrawal would be overridden in safety-critical scenarios without impairing vehicle control.

Authors: We agree that the current manuscript presents the framework at a conceptual level and lacks explicit mappings to specific regulatory provisions or exception analyses. The modular design is intended to support such adaptations, but the text does not currently demonstrate this through concrete examples. In the revised version, we will insert a dedicated subsection that provides explicit mappings: compatibility with GDPR Article 9 for biometric data processing, alignment with EU AI Act high-risk obligations for in-cabin monitoring, and mechanisms for consent override in safety-critical situations that respect legal exceptions (e.g., public interest and vital interests) while preserving vehicle control integrity. This addition will directly support the claim and address the identified gaps. revision: yes
Referee: [Risk analysis and failure mitigation strategy] The risk analysis and failure mitigation strategy (outlined in the final contribution) presents proactive incident response and accountability mechanisms at a high level but does not include case-specific mappings, compliance checklists, or evaluation criteria to verify that the safeguards prevent the very regulatory conflicts identified earlier in the review.

Authors: We acknowledge that the risk analysis and mitigation strategy is currently described at a high level without case-specific illustrations or verification tools. To strengthen this section, we will expand it to include concrete case mappings (e.g., biometric data breach or fairness violations in emotion detection), compliance checklists tied to each framework component, and proposed evaluation criteria such as transparency metrics and accountability audit protocols. These additions will demonstrate how the safeguards mitigate the regulatory conflicts noted in the review. revision: yes

Circularity Check

0 steps flagged

No circularity: framework is external synthesis with no self-referential reductions

full rationale

The paper performs a review of external instruments (GDPR, EU AI Act, IEEE guidelines) to identify gaps in applicability to multimodal in-cabin sensing, then presents a modular design framework as a direct translation of those high-level principles into components such as consent mechanisms and fairness-aware models. No equations, fitted parameters, self-citations, or uniqueness theorems appear in the provided text; the derivation chain consists of external review followed by independent synthesis rather than any step that reduces by construction to the paper's own inputs or prior outputs. The central proposal therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The paper rests on the assumption that existing high-level regulations are insufficient for DMS-specific risks and that a modular framework can translate them into actionable guidance without introducing new fitted parameters or unverified entities.

axioms (1)

domain assumption Overarching frameworks such as the GDPR, EU AI Act, and IEEE standards offer important guidance but lack the specificity required for addressing the unique risks posed by in-cabin sensing technologies.
Stated directly in the abstract as the motivation for the review and framework.

invented entities (1)

Modular ethical design framework for DMS no independent evidence
purpose: To translate high-level ethical principles into actionable design guidance including consent mechanisms, fairness-aware models, transparency tools, and emotional safeguards.
Newly proposed in the paper as the core contribution; no independent evidence or external validation provided.

pith-pipeline@v0.9.0 · 5588 in / 1591 out tokens · 44007 ms · 2026-05-08T04:47:34.123384+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

17 extracted references · 11 canonical work pages

[1]

Ac- cessed 10 April

Twitter says its image crops weren’t very biased, but is phasing them out anyhow.https://www.theverge.com/2021/5/19/ 22444372/twitter-image-crop-racial-gender-bias-research. Ac- cessed 10 April

2021
[2]

DATA, P.O.P.,

Autonomous vehicles: human factors issues and future research, in: Australasian Road Safety Conference, 1st, 2015, Gold Coast, Queensland, Australia. DATA, P.O.P.,

2015
[3]

European New Car Assessment Programme (Euro NCAP)

Euro ncap vision 2030: a safer future for mobility. European New Car Assessment Programme (Euro NCAP) . European Union Agency for Cybersecurity,

2030
[4]

Farooq, M.A., Kielty, P., Yao, W., Corcoran, P., 2025a

Advanced techniques and use cases. Farooq, M.A., Kielty, P., Yao, W., Corcoran, P., 2025a. Synadult: Multi- modal synthetic adult dataset generation via diffusion models and neuro- morphic event simulation for critical biometric applications. IEEE Access . Farooq, M.A., Shariff, W., Corcoran, P., 2025b. Thermvision: Exploring flux for synthesizing hyper-r...

work page doi:10.1145/3746027.3755448 2025
[5]

IEEE Access 12, 145955– 145967

Joint speech- text embeddings for multitask speech processing. IEEE Access 12, 145955– 145967. doi:10.1109/ACCESS.2024.3473743. Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y.,

work page doi:10.1109/access.2024.3473743 2024
[6]

arXiv preprint arXiv:2307.13008

Adap- tation of whisper models to child speech recognition. arXiv preprint arXiv:2307.13008 . Jain, R., Yiwere, M.Y., Bigioi, D., Corcoran, P., Cucu, H.,

work page arXiv
[7]

IEEE Access 10, 47628–47642

A text- to-speech pipeline, evaluation methodology, and initial fine-tuning results for child speech synthesis. IEEE Access 10, 47628–47642. doi:10.1109/ ACCESS.2022.3170836. Jambholkar, M.,

work page arXiv 2022
[8]

arXiv preprint arXiv:2409.17380

Tesla’s autopilot: Ethics and tragedy. arXiv preprint arXiv:2409.17380 . Johnson, N., Li, Y., Tang, F., Sarker, S.,

work page arXiv
[9]

Privacy and initial information in automated driving—evaluation of information demands and data sharing concerns, in: 2017 IEEE Intelligent Vehicles Symposium (IV), IEEE. pp. 541–546. Jung, J., Lim, S., Kim, B.K., Lee, S.,

2017
[10]

IEEE Access 11, 96363–96373

Neuromorphic driver monitoring systems: A proof-of-concept for yawn detection and seatbelt state detection using an event camera. IEEE Access 11, 96363–96373. doi:10.1109/ACCESS.2023.3312190. Koesdwiady, A., Soua, R., Karray, F., Kamel, M.S.,

work page doi:10.1109/access.2023.3312190 2023
[11]

arXiv preprint arXiv:2408.15388

Panoptic perception for autonomous driving: A survey. arXiv preprint arXiv:2408.15388 . Liu, Y.,

work page arXiv
[12]

Would i consent if it monitors me better? a technology acceptance com- parison of bci-based and unobtrusive driver monitoring systems, in: 2022 IEEE International Conference on Metrology for Extended Reality, Ar- tificial Intelligence and Neural Engineering (MetroXRAINE), IEEE. pp. 545–550. Regulation, P.,

2022
[13]

IEEE Access 11, 76964–76976

Real-time multi-task facial analytics with event cameras. IEEE Access 11, 76964–76976. doi:10.1109/ACCESS.2023.3297500. Ryan, C., Murphy, F., Mullins, M.,

work page doi:10.1109/access.2023.3297500 2023
[14]

IEEE Open Journal of Vehicular Technology 4, 836–848

Neuromorphic driver monitoring systems: A compu- tationally efficient proof-of-concept for driver distraction detection. IEEE Open Journal of Vehicular Technology 4, 836–848. doi:10.1109/OJVT. 2023.3325656. Shariff, W., Dilmaghani, M.S., Kielty, P., Moustafa, M., Lemley, J., Cor- coran, P.,

work page doi:10.1109/ojvt 2023
[15]

IEEE Access 12, 51275–51306

Event cameras in automotive sensing: A review. IEEE Access 12, 51275–51306. doi:10.1109/ACCESS.2024.3386032. Shelby, R., Rismani, S., Henne, K., Moon, A., Rostamzadeh, N., Nicholas, P., Yilla-Akbari, N., Gallegos, J., Smart, A., Garcia, E., et al.,

work page doi:10.1109/access.2024.3386032 2024
[16]

Sociotechnicalharmsofalgorithmicsystems: Scopingataxonomyforharm reduction, in: Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society, pp. 723–741. 45 Sherstinsky, A.,

2023
[17]

arXiv preprint arXiv:2103.02162

Predicting driver fatigue in auto- mated driving with explainability. arXiv preprint arXiv:2103.02162 . Zuboff, S.,

work page arXiv