Sadallah, M., Encelle, B., Mille, A., Guillet, V., & Virgo, J. (2026). Partenariat épistémique et recherche participative : le projet COOBRA comme démonstrateur méthodologique. In 14ème édition de la conférence sur les recherches pluridisciplinaires pour l’autonomie des personnes en situation de handicap (Handicap '26). Aubervilliers & Paris: IFRATH. 10-12 juin 2026.
Summary: This article highlights the persistent exclusion of visually impaired people from scientific pathways due to inaccessible technical documents and unequal collaboration with sighted people. We propose an epistemic partnership approach based on the co-construction of knowledge, through an accessible, collaborative, and egalitarian methodology implemented within an ANR project on digital accessibility. Our approach demonstrates the possibility of conducting rigorous fully co-constructed research while transforming encountered difficulties into drivers of technical innovation.

Sadallah, M., Encelle, B., Mille, A. & Guillet, V. (2026). Partenariat épistémique et recherche participative : le projet COOBRA comme démonstrateur méthodologique. In 14ème édition de la conférence sur les recherches pluridisciplinaires pour l’autonomie des personnes en situation de handicap (Handicap '26). Aubervilliers & Paris: IFRATH. 10-12 juin 2026.
Summary: This article challenges the idea that audio can replace braille in digital accessibility, particularly in scientific and technical fields where navigation and verification requirements make braille indispensable. We propose the Anchor and Flow model, in which braille provides cognitive stability and information verification, while audio and AI support exploration and fluid interaction. In response to the risks of generative AI hallucinations, we argue that braille is an essential tool for epistemic auditing and cognitive sovereignty in the algorithmic era.

Sadallah, M., & Encelle, B. (2026). Beyond Additive Design: An Empirical Taxonomy of Multimodal STEM Accessibility Systems. In Extended abstracts of the CHI conference on human factors in computing systems (CHI '26). Barcelona, Spain: ACM, April 13-17, 2026.
Summary: An empirical taxonomy of multimodal systems for accessing STEM content by people who are blind or visually impaired, based on an analysis of 66 systems. Three design approaches are identified (additive, augmentative, integrative), and a decoupling between technical performance and cognitive load is highlighted.

• Sadallah, M., & Encelle, B. (2026). The verification gap in AI accessibility: A meta-analysis of architectural paradigms and agency deficits. In Proceedings of the 23rd International Web for All Conference (W4A'26). Dubai, United Arab Emirates: ACM.
  Summary: A meta-analysis of 17 AI-based accessibility systems (2019–2025) that introduces the Verification-Cognition Balance Index (VCBI). We show that transformer architectures exhibit a higher risk density than traditional computer vision and advocate for a paradigm shift toward user-verifiability.

Sadallah, M., & Encelle, B. (2026). The Technical Capability Fallacy: A Case for Pedagogy-Centered Design in Assistive STEM Education. In 18th International Conference on Computer Supported Education (CSEDU 2026). Benidorm, Spain, May 18-20, 2026.
Résumé : An analysis of 25 empirical studies showing that assistive tools for STEM learning among people with visual impairments fail in real-world settings because they are designed based on technical metrics that do not predict educational success. Proposal of the FOCAL framework (Framework for Objective-Context ALignment).

SLR1 – Access to 2D Structures

• Sadallah, M., & Encelle, B. (2026). Systematic Review Protocol: Interactive Digital Technologies for Non-Visual Access to 2D STEM Content by Blind and Low-Vision Individuals. OSF : https://doi.org/10.17605/OSF.IO/TJ5QZ ; Zenodo : https://doi.org/10.5281/zenodo.19461731.
  Summary: Content: This protocol details the methodology of the project’s first systematic review. It incorporates three complementary frameworks (PRISMA 2020, SPIDER, MMAT 2018) as well as cognitive load theory and the Technology Acceptance Model (TAM/UTAUT). It provides five original methodological tools: an integration model, an extraction taxonomy (947 controlled terms), dual extraction templates (35 quick fields / 654 full fields), an automated validation infrastructure, and a structured synthesis plan. Inter-rater reliability is excellent (ICC = 0.92).
  Status: Pre-registered on OSF; protocol version 2.0 (February 2026). All materials are available under open access (CC-BY, MIT).

SLR2 – Synchronous collaboration between sighted and visually impaired individuals – in progress, results expected in the second half of 2026

• Practices and barriers faced by people with visual impairments when accessing scientific documents – data analysis (Spring 2026)

• November 2025 – Project presentation at the COOBRA plenary meetings – slides (accessible format)
• January 2026 – “State of the Art and Needs” workshop – slides
• Spring 2026 – LIRIS seminar (to be announced)
• 2026 – HANDICAP Conference (submission in progress)

• Deliverable D1 – SLR1 PRISMA Protocol (co-developed, accessible) – Tagged PDF | HTML
• Deliverable D2 – Survey Questionnaire and Accessibility Report – HTML | Tagged PDF | Braille upon request
• Deliverable D3 – Functional Specifications for Collaborative Prototypes (coming soon)

• COOBRA Project Overview (2 pages) – text | audio | Braille
• Short presentation (5 minutes) for non-specialists – video with subtitles | slides
• Scientific poster (accessible format) – PDF