Inspiration Space debris is one of the greatest threats to our orbital future. With over 36,000 tracked objects (and millions of smaller fragments) hurtling around Earth at speeds up to 28,000 km/h, we're inching closer to Kessler Syndrome — a cascading collision scenario that could render low-Earth orbit unusable for generations, crippling satellite communications, GPS, weather forecasting, and future Moon/Mars missions. Current tracking relies on government radars with blind spots, especially in underserved regions. Yet many debris pieces (and satellites) are visible to the naked eye during twilight as bright streaks. We were inspired by citizen science successes in other fields (like marine debris tracking apps) and asked: What if everyday people with smartphones could become the world's largest distributed space surveillance network? DebrisGuard turns this vision into reality — democratizing space safety and helping keep orbits clean for humanity's multi-planetary future.
What it does DebrisGuard is an AI-powered mobile web app that transforms smartphones into citizen satellite spotters. Users point their camera at the twilight sky, capture photos/videos of bright streaks, and the AI instantly analyzes them to classify known satellites vs. potential new debris fragments. Verified anomalous sightings feed into a real-time global heatmap of risk zones, flagging untracked threats. The app gamifies participation with badges and leaderboards, educates users about orbital mechanics, and shares open data/alerts with satellite operators and researchers — creating proactive, crowdsourced early warnings to prevent collisions and protect critical space infrastructure.
How I built it I built a beginner-friendly prototype using low-code/no-code tools for speed: FlutterFlow for the mobile-friendly UI, integrated phone camera + basic AR overlays (via 8th Wall webAR), and Firebase/Supabase for real-time database and sightings storage. The core AI magic comes from multimodal models (Grok + Gemini APIs) — we engineered prompts to analyze images for streak characteristics (speed, direction, brightness) and cross-reference against public TLE data from Celestrak. We added a simple 3D globe visualization with Globe.GL to display heatmaps. Everything runs edge-first for privacy, with optional cloud boosts.
Challenges we ran into Time was tight (same-day hackathon style!), so balancing AI accuracy with on-device speed was tough — early models confused planes/stars with debris. Integrating real-time orbital data APIs had rate limits and CORS issues. We also had to ensure privacy (anonymized GPS) while making the app engaging.
Accomplishments that we're proud of We created the first citizen-science smartphone pipeline focused on proactive debris discovery using AI vision — not just known satellite tracking. The visual demo (snap → classify → heatmap update) looks impressive, and the social good angle (protecting global access to space) feels truly impactful. We disclosed AI usage transparently and built something scalable with real-world potential.
What we learned Prompt engineering is an art — small tweaks dramatically improved classification. Citizen science thrives on gamification and education. Most importantly, space problems are global; solutions need inclusive, low-barrier tools to succeed.
What's next for DebrisGuard Post-hackathon, we plan to: refine the AI with more training data, add community-verified alerts via push notifications, partner with organizations like ESA/NASA for data validation, expand to iOS/Android native apps, and integrate delay-tolerant networking for future interplanetary use (e.g., Mars orbital debris spotting). Ultimately, we want DebrisGuard to become a global movement — millions of "Space Guardians" safeguarding orbits for everyone.
Built With
- firebase
- flutterflow
- geminiapi
- groq
- javascript
- webar
Log in or sign up for Devpost to join the conversation.