Inspiration Landslides are a growing global crisis, especially in tropical regions like Southeast Asia where extreme rainfall causes devastating disasters every year. In early 2026, record-high fatalities highlighted the urgent need for accessible early-warning systems. This inspired us to create RaiHalai (“Landslide Warning”), a platform that transforms smartphones into AI-powered landslide detection tools for vulnerable communities. What It Does RaiHalai provides real-time landslide risk analysis using AI, IoT, and geospatial data. The platform calculates risk scores between: 0–1000\text{--}1000–100 based on rainfall, terrain slope, and soil saturation. Users can scan terrain using their smartphone camera to detect cracks, erosion, and instability indicators with an estimated accuracy of 78–95%. The system also integrates IoT sensors to monitor soil moisture, rainfall intensity, and micro-vibrations in real time. RaiHalai sends localized emergency alerts in English, Tetum, Indonesian, and Portuguese through a Progressive Web App (PWA) that continues working even with unstable internet connections. How We Built It We built RaiHalai using: React TypeScript Tailwind CSS Supabase Edge Functions The platform uses AI image analysis, real-time risk computation, Service Workers, and offline caching to ensure fast and resilient performance across Android and iOS devices. Challenges & Learnings One of our biggest challenges was maintaining reliable push notifications and system stability during external API outages. To solve this, we implemented a Graceful Fallback Mechanism that switches to simulated environmental data whenever third-party services fail. Through this project, we learned the importance of resilience, offline accessibility, and human-centered design when building disaster-response technology. What’s Next Next, we plan to integrate: Edge AI for offline image analysis Offline hazard maps Community crowdsourcing features for reporting terrain cracks and environmental hazards Our goal is to make RaiHalai a scalable and accessible disaster resilience platform for at-risk communities worldwide.

Built With

  • apis
  • medo
Share this project:

Updates