SafeSpot

• 

  Selection Home Page

• 

  Earthquake

• 

• 

• 

  Fire

• 

• 

  Learning Panel

💡 Inspiration 💡

We believe that every second counts in an emergency and that technology can bridge the gap between panic and survival. When disasters strike - earthquakes, fires, floods, tornados - people often make dangerous decisions due to lack of preparedness knowledge or visual impairments. We were inspired by the lack of AR applications addressing real emergency situations and wanted to create something that could genuinely save lives by providing instant, location-specific safety guidance when it matters most.

References: https://www.ready.gov/earthquakes, https://www.redcross.org/get-help/how-to-prepare-for-emergencies

⚙️ What it does ⚙️

SafeSpot is an AR-powered emergency guidance app that transforms your phone or iPad into a life-saving safety assistant. Users select their emergency type (earthquake, fire, flood, or tornado) and point their device around their surroundings. The app instantly overlays color-coded safety instructions directly onto real-world objects in real-time.

During an earthquake, SafeSpot might highlight a sturdy table with green bounding boxes and "GET UNDER" text, or warn about a bookshelf with red boxes and "STAY AWAY." The app works entirely offline, ensuring it remains functional when networks fail during disasters - critical for disaster-prone regions where internet connectivity is unreliable.

Each detection is powered by AI that understands hazard-specific safety protocols, providing personalized guidance based on your exact environment rather than generic safety tips.

🏗️ How we built it 🏗️

For the AR interface, we used RoomPlan API and RealityKit. For the backend safety logic, we implemented custom hazard detection algorithms and offline AI models.

🟣 RoomPlan API 🟣

We leveraged Apple's RoomPlan API to achieve real-time object detection without requiring pre-scanning. This allows users to immediately start receiving safety guidance as soon as they open the app - crucial during high-stress emergency situations. The AI classifies detected objects based on the selected emergency type - what's safe during an earthquake (sturdy furniture) differs from what's safe during a fire (clear exit paths). All processing happens on-device for offline functionality. We use Apple's RoomPlan API which utilizes computer vision models.

🔴 RealityKit, AR, and Graphics 🔴

SafeSpot uses RealityKit's advanced 3D rendering to create immersive safety zones that completely enclose detected areas with dynamic mesh geometries. Rather than flat overlays, we generate volumetric 3D zones using custom PBR materials with emissive properties that create a distinctive glowy effect

⚫ User Experience ⚫

The app features a simple grid-based hazard selection screen and intuitive AR guidance with color-coded bounding boxes and clear text instructions. We prioritized accessibility with high contrast modes and large text for users with visual impairments, and designed for clarity during high-stress situations.

🚩 Challenges we ran into

We discovered that creating accurate real-time object detection without pre-scanning was technically demanding, especially while maintaining smooth performance on mobile devices. Balancing the visual complexity of AR overlays with the need for instant clarity during emergencies required extensive user experience testing. We also had to research and validate safety protocols for different disaster types to ensure our guidance was accurate and potentially life-saving.

🥇 Accomplishments that we're proud of

Successfully implemented real-time AR object detection with RoomPlan API that works without pre-scanning. Created an intuitive, accessible interface that functions offline. Developed the first AR application specifically designed for emergency preparedness - filling a critical gap in the App Store. Built comprehensive safety protocols for multiple disaster types with clear, actionable guidance.

📚 What we learned

We learned that emergency preparedness technology has been largely overlooked in the AR space, despite its life-saving potential. We gained deep technical knowledge about real-time object detection, AR interface design, and the importance of offline functionality for critical applications. Most importantly, we discovered how technology can truly serve humanity's most fundamental need: safety.

⏳ What's next for SafeSpot?

We plan to expand support for additional emergency types (wildfires, chemical spills, active threats), integrate with emergency services for real-time alerts, and add multi-language support for global accessibility. We're also exploring partnerships with emergency response organizations and considering Apple Watch integration for hands-free guidance. Long-term, we envision SafeSpot becoming a standard emergency preparedness tool in disaster-prone regions worldwide.

Built With

• ar/vr
• computer-vision
• graphics
• realitykit
• swift