ARise: Augmented Reality for Heritage Resilience
- ARise is an augmented reality mobile application that enhances cultural heritage resilience through immersive, user-centered engagement with local sites.
- It integrates crowdsourced hazard reports, social media sentiment analysis, and AI-generated artworks to convert abstract climate data into interactive AR content.
- Designed through a co-creative process across five European regions, ARise fosters community awareness, adaptive behaviors, and cultural sustainability.
ARise is an augmented reality mobile application designed to improve cultural heritage resilience by enhancing public engagement with local cultural sites while raising awareness of localized climate impacts and risks affecting heritage. It was developed through a user-centered, Design Thinking-inspired, co-creative methodology involving stakeholders from five European regions, and it integrates a Crowdsourcing Chatbot, a Social Media Data Analysis tool, and an AI-based Artwork Generation module into a modular AR mobile platform supporting both geolocated onsite experiences and plane-anchored offsite exploration (Urbanelli et al., 3 Nov 2025).
1. Conceptual scope and resilience framing
ARise is situated in a problem context in which cultural heritage is increasingly exposed to climate change and environmental hazards that degrade sites, interrupt access, and erode the social fabric sustained by heritage. The relevant hazards include floods, fires, storms, heat stress, and ecosystem changes, and the threatened assets span both tangible assets such as buildings, monuments, and landscapes and intangible values such as identity, memory, and practices (Urbanelli et al., 3 Nov 2025).
Within this framing, resilience encompasses awareness, preparedness, adaptive behaviors, and community engagement. ARise operationalizes that framing qualitatively rather than through formal indices: it translates abstract risk and climate data into place-based, emotionally engaging AR content; enables citizens to contribute situational intelligence via crowdsourced reports; provides professionals and the public with sentiment and relevance indicators; and supports education and adaptive behavior through interactive exploration of environmental indicators on local terrains (Urbanelli et al., 3 Nov 2025). No formal models or equations are presented for risk or resilience; the approach is explicitly qualitative and design-led.
The system’s stated goals are to engage the public with local cultural sites through immersive, emotionally resonant experiences, raise awareness of localized climate impacts and risks affecting heritage, and support education, cultural sustainability, and climate adaptation by making environmental and social data accessible and actionable. Relative to prior AR work in cultural heritage, the prototype is distinguished by a unified three-source data ecosystem, the combination of geolocated onsite and plane-anchored offsite experiences, a co-creative multi-region methodology, and gamified engagement (Urbanelli et al., 3 Nov 2025).
2. Co-creative development and regional embedding
ARise was co-created within a broader resilience ecosystem through a user-centered process with stakeholders from Italy, Spain, Greece, Germany, and Croatia. The stakeholder contexts comprised three UNESCO sites, one FAO GIAHS site, and one National Park, and the stated purpose of this multi-region configuration was to ensure relevance and adaptability across heterogeneous heritage environments (Urbanelli et al., 3 Nov 2025).
The co-design activities were structured in five phases: collecting data and knowledge; understanding risks and impacts; developing strategies and solutions; disseminating knowledge to targeted stakeholders; and informing people to promote awareness and engagement. Requirements gathering was organized into four categories—UX/UI, Functional, Content, and Deployment—and the selection criteria for Points of Interest and datasets were established jointly with local stakeholders. The chatbot data schema was also expanded with stakeholder-prioritized measurement types and risk indicators (Urbanelli et al., 3 Nov 2025).
This development model is significant because the paper positions ARise not as a generic AR demonstrator, but as a regionally grounded system whose content variations are tied to locally selected Points of Interest, region-specific hazards, and top-reviewed sites per region. The paper situates this design stance relative to prior reviews of AR in cultural heritage, XR frameworks in cultural heritage, AI in cultural heritage, co-design for AR heritage, and Design Thinking literature (Urbanelli et al., 3 Nov 2025).
3. Data ecosystem and modular architecture
ARise is built as a modular AR mobile platform that merges outputs from three tools. The architecture centers on a three-source pipeline in which crowdsourced hazard intelligence, social media analytics, and generative AI outputs are transformed into AR content for cultural heritage interpretation and climate-risk awareness (Urbanelli et al., 3 Nov 2025).
| Component | Role | Main outputs |
|---|---|---|
| Crowdsourcing Chatbot | Collects real-time, geolocated hazard and site reports | GPS coordinates, hazard type, descriptions, multimedia, risk-related fields |
| Social Media Data Analysis tool | Monitors public perception and relevance of selected sites | Sentiment score, number of reviews, importance score, image-derived descriptions |
| AI-based Artwork Generation module | Produces artworks reflecting public sentiment associated with each site | AI-generated “AR gallery” images |
The Crowdsourcing Chatbot is Telegram-based. It captures GPS coordinates, hazard type such as fire, flood, or storm, textual descriptions, photos, videos, voice messages, plus stakeholder-requested measurement types, impact indicators, and risk assessment elements. Users submit reports through the chatbot; validated entries are stored and made available to ARise for geolocated visualization at Points of Interest (Urbanelli et al., 3 Nov 2025).
The Social Media Data Analysis tool monitors Google Maps reviews and images for stakeholder-identified Points of Interest. Its methods include NLP-based sentiment analysis assigning a numerical score to each review, an importance score combining sentiment with the number of reviews, and computer vision-style processing to extract descriptions from Google Maps images. These outputs feed both the AR app and the AI artwork generation pipeline (Urbanelli et al., 3 Nov 2025).
The AI-based Artwork Generation module uses Stable Diffusion v1.5, identified as a latent diffusion model per Rombach et al. and pretrained on LAION-5B by CompVis/RunwayML. Periodically, for each region, it retrieves the top five Points of Interest by review volume from the SMDA tool, including sentiment scores and photographs, and uses sentiment as a prompt condition to transform site photos into new artistic images, producing an “AR gallery” content set (Urbanelli et al., 3 Nov 2025).
4. Onsite and offsite AR interaction design
ARise divides interaction into OnSite and OffSite modes. In onsite mode, GPS and plane detection are used to place 3D hazard models and Point-of-Interest markers at their true coordinates. Tapping a model opens overlays containing report details such as type, distance, description, and images, or SMDA summaries including review count, sentiment score, and importance (Urbanelli et al., 3 Nov 2025).
Offsite mode uses plane recognition to anchor content anywhere. This enables three classes of experience: interactive 3D maps visualizing social datasets such as sentiment and importance per site; immersive AR galleries displaying AI-generated artworks tied to local heritage and sentiment; and climate impact visualizations on 3D terrain maps where users explore changes in indicators such as temperature, water levels, and vegetation coverage. User interactions update the visualization to convey potential effects (Urbanelli et al., 3 Nov 2025).
The interaction model is explicitly designed around immersion, emotional engagement, and personalization. A modular navigation flow begins with authentication and a personalized home screen, after which users switch between OnSite and OffSite. Interactions include tapping geolocated models, manipulating 3D maps, entering an AR “portal” into an artworks gallery, and exploring indicator UIs. Emotional engagement is supported through sentiment-conditioned artworks, overlays that translate abstract data into place-based stories, and a gamified system that tracks interactions and rewards exploration (Urbanelli et al., 3 Nov 2025).
The representational layer includes hazard icons and models, terrain models reconstructed from satellite imagery via height maps, interactive graphs, and AI-generated images displayed in AR galleries. Storytelling ties citizen reports and public sentiment to specific places, while authentication supports personalized progress tracking and localized content delivery for selected Points of Interest and regions (Urbanelli et al., 3 Nov 2025).
5. Technical implementation, evaluation status, and limitations
The application is implemented in Unity with AR Foundation for cross-platform AR and real-time spatial tracking. It is designed for modern smartphones and requires GPS, camera, and AR-capable hardware. The system was internally tested for stable plane detection and real-time rendering of 3D assets and overlays, and the content selection and trigger mechanisms were tuned according to dataset type, specifically GPS-based versus plane-detected content (Urbanelli et al., 3 Nov 2025).
Content management is periodic and modular. SMDA outputs are periodically retrieved to update Point-of-Interest datasets; the AI module periodically generates artworks based on top-reviewed sites per region; and crowdsourced reports flow from Telegram to ARise’s geolocated visualizations. The cited data sources are Google Maps reviews and images, citizen chatbot reports, and satellite imagery for terrain reconstruction. A demo video is hosted on Zenodo (Urbanelli et al., 3 Nov 2025).
The current status is that a fully functional prototype has been developed and internally tested, while formal end-user evaluation remains forthcoming. The paper states that future work will conduct user tests to assess usability and effectiveness in meeting end-user needs and to refine features accordingly. It also notes that no formal evaluation framework is yet provided, while identifying typical dimensions for the domain such as engagement, learning outcomes, attitude or behavioral intentions, usability and accessibility, and emotional resonance (Urbanelli et al., 3 Nov 2025).
The paper also identifies several limitations. There is no formal end-user testing yet; privacy and ethics details are not elaborated; dependence on Google Maps reviews limits the breadth of social data; and environmental visualizations present indicator effects without coupling to predictive climate models. The discussion of ethics is therefore prospective: it states that a production system would need clear user consent and purpose limitation, anonymization or pseudonymization of personally identifying information, moderation workflows for sensitive or inappropriate content and hazard verification, compliance with EU regulations such as GDPR, and bias mitigation strategies for sentiment analysis and image processing (Urbanelli et al., 3 Nov 2025).
6. Comparative positioning and acronym disambiguation
Within cultural heritage and climate-awareness tooling, ARise is positioned as a prototype whose novelty lies in a unified, multi-source pipeline rendered coherently in AR; the combination of onsite geolocated overlays and offsite plane-anchored experiences; an emotion-centric design that pairs data with AI-generated cultural imagery; and a co-creative, multi-region methodology spanning UNESCO, FAO GIAHS, and National Park contexts (Urbanelli et al., 3 Nov 2025).
A frequent source of confusion is that the acronym “ARISE” is used by numerous unrelated arXiv works. In 2025–2026 it also denotes a framework for knowledge-augmented reasoning via risk-adaptive search (Zhang et al., 15 Apr 2025), a system for automating RISC-V instruction-set extension (Hager-Clukas et al., 11 Aug 2025), an adaptive resolution-aware metric for test-time scaling evaluation in large reasoning models (Yin et al., 7 Oct 2025), an automated scholarly survey-generation engine (Wang et al., 21 Nov 2025), and a repository-level graph toolset for agentic fault localization and program repair (Seddik et al., 4 May 2026). The same acronym also appears in reinforcement learning (M et al., 2 Jan 2026, Li et al., 17 Mar 2026), autonomous-driving scenario generation (Poddubnyy et al., 21 Jan 2026), text classification (M. et al., 9 Feb 2025), speech enhancement (Shen et al., 28 May 2025), graph anomaly detection (Duan et al., 2022), granular matter experiments on the International Space Station (Steinpilz et al., 2019), and semi-parametric financial time-series modeling (Zhang et al., 2021). In the cultural-heritage literature, however, “ARise” refers specifically to the augmented reality mobile application for cultural heritage resilience described in (Urbanelli et al., 3 Nov 2025).