Rakshak-AI: Autonomous Road-Accident Detection and Emergency Triage Platform
A pan-India, edge-powered ecosystem that autonomously detects severe road accidents using smartphone telemetry, dispatching GPS coords to ambulances and activating traffic preemption.
Stall Participants & Presenters
Overview
India faces severe casualties and loss of life during road incidents due to delayed emergency responses and missed “Golden Hours” caused by urban gridlock and bystander reliance.
Rakshak-AI is a pan-India, edge-powered accident detection and emergency triage platform designed to address these bottlenecks. By eliminating human reporting delays, Rakshak-AI dynamically engineers rescue routes, coordinates emergency services, and works to reduce nationwide traffic mortalities, directly aligning with SDG 3 (Good Health and Well-Being) and SDG 9 (Industry, Innovation, and Infrastructure).
Edge Telemetry & Intelligent Triage
The system operates directly on mobile edge devices using real-time sensor processing:
- Autonomous Crash Detection: Samples smartphone IMU telemetry at 50Hz, processing accelerometer and gyroscope vectors through an edge-optimized TensorFlow Lite model trained to distinguish actual high-impact collisions from potholes and sudden stops.
- Instant Dispatch & Geolocation: Automatically captures and streams precise GPS coordinates to the nearest emergency dispatch center and active ambulance fleet.
- C-V2X “Green Wave” Preemption: Interfaces with smart traffic management systems to orchestrate green corridors for approaching emergency vehicles.
- Trauma Center Triage Integration: Pre-alerts receiving trauma centers with ABHA-linked patient vitals and incoming injury telemetry to prepare critical care units before arrival.
- Privacy & Compliance: Adheres to DPDP standards with zero persistent storage of non-incident telemetry.
Architecture & Technology Stack
- Mobile Client: Flutter and native Kotlin modules for low-power background telemetry processing, hardware sensor listener relays, and lock-screen override.
- Edge AI: TensorFlow Lite neural model evaluating IMU telemetry windows at 50Hz for immediate crash classification.
- Backend & Ingestion: Node.js backend with Socket.io and WebSockets for low-latency dispatch streaming and real-time fleet synchronization.
- Spatial Data & Caching: Redis for fast real-time coordinate caching and spatial queuing, paired with PostgreSQL/PostGIS for geographic radius search and route analysis.
- Emergency Protocols: C-V2X protocol emulation for signal preemption and webhook integrations with trauma medical portals.

