Abstract
The project evaluates the application of extended reality and artificial intelligence to humanitarian emergency readiness and response simulation exercises within the EU-funded CORTEX2 project. The objective was to determine the operational relevance of XR-supported immersive training for cooperation in crisis management, using a proof-of-concept developed by Nuwa in collaboration with Action Contre la Faim with platform components developed by Alcatel Lucent Enterprise, DFKI, Linagora, CEA, and other stakeholders from within the EU-funded CORTEX2 project. Three pilots were implemented: (1) an arrival briefing introducing key concepts, (2) an alert and response strategy exercise in collaborative team settings, and (3) a response implementation simulation engaging participants with AI avatars in immersive environments. Evaluation was conducted with Action Contre la Faim's emergency roster personnel through structured exercises and post-simulation surveys. Results indicate limited added value of XR for theoretical briefings, moderate value for collaborative planning, and highest potential for response implementation, particularly for practicing soft-skills, decision-making, and negotiation under pressure. Usability was constrained by interface complexity and speech recognition issues, but participants identified XR and AI as valuable for experiential training when applied selectively. Findings recommend targeted deployment of XR in defined phases of emergency response training, streamlined user interface design, and dual support for VR and desktop access. The research contributes to the design of technology-mediated humanitarian training by aligning immersive methods with specific operational needs.
Keywords
Humanitarian, Cooperation, Extended Reality, Artificial Intelligence, Disaster Risk Reduction
Introduction
When faced with emergency and crisis situations, the ability to follow protocols calmly whilst under extreme pressure is paramount, and crisis response teams undergo intensive training and continuous upskilling to cope and respond professionally. Additionally, it is difficult to replicate an emergency or humanitarian crisis in the real world, as this requires significant time and resources. Conventional methodologies for training and upskilling can be costly, inefficient, and logistically challenging. The XRisis project addresses these challenges by leveraging real-time collaborative communication functionality to create inclusive, engaging, and easily repeatable simulated virtual crisis environments without the associated danger encountered in the real world.
Platform Architecture and CORTEX2 Integration
The XRisis platform architecture integrated four distinct CORTEX2 enabling technologies within a Unity-based multi-user environment deployed on European cloud infrastructure. Rainbow CPaaS from Alcatel Lucent Enterprise provided secure real-time communication backbone supporting voice and video across all participant endpoints through Unity C# SDK integration. DFKI's Video Call Alternative Appearance enabled avatar-based video conferencing preserving presence whilst protecting privacy. CEA's Conversational Virtual Agent powered AI dialogue characters capable of domain-specific knowledge retrieval and contextual conversation. Linagora's summarisation agent generated automatic meeting summaries from call recordings supporting post-exercise analysis and documentation.
The technical stack coordinated multiple components: Unity server managing multi-player engine synchronisation, WebSocket server distributing environmental state across participant endpoints, web application providing facilitator control interfaces for scenario progression and inject management, and integration layers connecting CORTEX2 services with platform core functionality. Infrastructure deployed exclusively on EU-based cloud providers in Germany and France ensuring GDPR compliance and data sovereignty alignment with humanitarian sector requirements.
Three-Pilot Implementation Structure
Pilot 1: Arrival Briefing introduced participants individually to emergency management concepts through AI mentor "Maud" powered by CEA's Conversational Virtual Agent. Participants navigated virtual environments whilst engaging natural language dialogue about emergency classification, management cycle phases, and organisational coordination mechanisms. Learning objectives centred on knowledge recall and conceptual understanding foundational to subsequent application-focused pilots.
Pilot 2: Alert and Response Strategy positioned four-person teams in collaborative coordination office environments responding to flooding emergency scenario. Teams received simulated emails, news reports, and facilitator communication via Rainbow CPaaS whilst developing Emergency Preparedness Response Plans using interactive strategy design tools. Learning objectives emphasised analytical skills, collaborative decision-making, and effective communication under pressure.
Pilot 3: Implementation Simulation engaged individual participants in role-specific scenarios requiring stakeholder negotiation with AI-powered characters. Participants confronted implementation challenges including access negotiation, logistics coordination, community relationship management, and inter-agency collaboration through conversational interactions adapting to participant communication approaches. Learning objectives focused on soft skills including negotiation, cultural sensitivity, relationship building, and adaptive problem-solving.
Validation Methodology
Evaluation employed mixed-methods design combining System Usability Scale (standardised ten-item psychometric instrument), component-specific added value ratings (five-point Likert scales assessing training effectiveness), overall satisfaction measurement, and qualitative debrief analysis. Eight participants from Action Contre la Faim's emergency roster participated as two four-person teams on 14 May 2025 in Paris. Participant profiles included varied emergency deployment experience (five of eight had completed field deployments) and simulation exercise familiarity (six of eight had participated in three or more previous exercises). Data collection included written surveys administered post-exercise and verbal group debriefs facilitated separately with participants and project team members.
Results and Analysis
System Usability Scale yielded average 59% with distribution showing six of ten users scoring above 51% and two above 71%, indicating acceptable usability for motivated professional users whilst confirming refinement requirements. Component added value ratings showed significant differentiation: soft skills practice 4.2 out of 5, collaborative teamwork 3.6 out of 5, response strategy tool 3.4 out of 5, facilitator debrief 3.3 out of 5, and AI briefing 3.2 out of 5, with overall average 3.5 out of 5 (70% added value). User satisfaction averaged 3.3 out of 5 (66%) with 50% providing maximum ratings suggesting strong appeal for specific user profiles.
Qualitative analysis identified implementation simulation as uniquely valuable for soft skills development, collaborative planning as context-dependent with value for distributed teams but limited benefit for co-located participants, and theoretical briefing as poorly matched to XR delivery compared to conventional e-learning. Technical challenges included speech recognition failures for non-native speakers, interface complexity creating usability barriers, and unnecessarily elaborate environments generating cognitive distraction.
Discussion and Implications
Results demonstrate that extended reality delivers differential value across humanitarian training applications, with implementation phase simulation providing unique capabilities for soft skills practice whilst theoretical knowledge transfer shows minimal incremental benefit from immersive delivery. This finding informs selective deployment strategy prioritising high-value applications rather than comprehensive platform coverage. The usability score of 59% reflects genuine interface complexity requiring simplification whilst remaining within acceptable range for complex professional tools during initial deployment, suggesting refinement rather than fundamental redesign will achieve commercial viability thresholds. Desktop and VR headset dual support proved essential: participants using desktops reported adequate value without expensive hardware requirements, expanding market addressability whilst maintaining immersion enhancement options for organisations with existing VR investments.
The validation demonstrates that rigorous evaluation with operational users generates essential evidence for deployment decisions, challenging assumptions about XR universally enhancing training whilst providing concrete assessment of when, where, and how immersive technologies deliver proportional value justifying investment.
Conclusion
The XRisis evaluation validates extended reality as valuable for selective humanitarian training applications, particularly implementation phase soft skills development, whilst identifying clear limitations for theoretical knowledge transfer and context-dependent benefits for collaborative planning. The research contributes evidence-based assessment methodology and component-specific effectiveness data informing deployment strategy for humanitarian and potentially broader crisis management training applications. Continued development will focus on interface refinement, speech recognition improvement, and commercial platform evolution incorporating validation findings whilst maintaining core capabilities demonstrating highest value.
Acknowledgements
The authors acknowledge Action Contre la Faim's Emergency Readiness and Response Unit for partnership and validation access, CORTEX2 consortium members for enabling technology provision and technical support, and participants who contributed time and expertise to rigorous evaluation enabling evidence-based platform development.
Funding Statement
This work received funding from the European Union's Horizon Europe research and innovation programme under grant agreement number 101070192 (CORTEX2 - COoperative Real-Time EXperiences with EXtended reality). Views and opinions expressed are those of the authors only and do not necessarily reflect those of the European Union. Neither the European Union nor the granting authority can be held responsible for them.
Related
Partners
Industries
Products
Technologies
Related Portfolio Projects
