CrashLens is a real-time crash detection and emergency coordination platform that bridges the moment of impact to the people who need to respond — in seconds, not minutes.
A Raspberry Pi 5 edge device reads live IMU, GPS, camera, and LTE data. When a crash is detected, the system instantly:
- Captures a 15-second pre-crash + 30-second post-crash video clip
- Uploads GPS coordinates, speed, and impact severity to the cloud
- Notifies the insurance dashboard with AI forensic analysis of the incident
"From sensor to responder in under 3 seconds."
CrashLens has been tested and verified on real hardware:
| Component | Status |
|---|---|
| Camera Module 3 NoIR (1080p30) | Proven |
| GPS Module (NEO-6M, UART) | Proven |
| IMU Crash Detection (100Hz, 2.5G threshold) | Proven |
| Ring Buffer Recording (15s pre + 30s post) | Proven |
| H264 → MP4 Conversion (ffmpeg) | Proven |
| Cloud Upload (metadata + video to R2) | Proven |
| Heartbeat Monitoring | Proven |
| Systemd Auto-Start on Boot | Proven |
| Insurance Dashboard (8 pages) | Proven |
+------------------------------------------------------------------+
| EDGE DEVICE |
| Raspberry Pi 5 · IMU/GPS/Camera/LTE |
| Real-time crash detection · AI inference on-device |
+----------------------------+-------------------------------------+
|
LTE/HTTPS
|
v
+------------------------------------------------------------------+
| CLOUD BACKEND |
| Cloudflare Workers + Hono.js (edge API) |
| Supabase (Postgres + Auth) |
| Cloudflare R2 (video evidence storage) |
| Multi-stakeholder routing |
+--------------------+--------------------------+------------------+
| |
+-------+------+ +--------+-----------+
| Flutter App | | Next.js Dashboard |
| Driver + | | Insurance / Fleet |
| Responder | | Management |
+--------------+ +--------------------+
| Layer | Technology |
|---|---|
| Edge Device | Raspberry Pi 5, Python, OpenCV, TensorFlow Lite |
| Connectivity | LTE (SIM7600G), GPS (NEO-6M), 6-axis IMU |
| Cloud API | Cloudflare Workers + Hono.js (TypeScript) |
| Database | Supabase (PostgreSQL + Row Level Security) |
| Video Storage | Cloudflare R2 |
| Mobile | Flutter (iOS + Android) |
| Dashboard | Next.js 14, TypeScript, Tailwind CSS |
| AI | On-device crash inference, cloud forensics pipeline |
CrashLens is a live, deployed commercial product. The platform is currently in active development and pilot testing.
Visit: crashlens.org
Elia Ghazal — Co-Founder & Lead Engineer GitHub · [email protected]
Commercial product — source code is proprietary. This repository is a public showcase only.