DRISHTI transforms ordinary smartphones into intelligent, cooperative safety devices by leveraging on-device sensors (GPS, NavIC, IMU, Camera), AI-based obstacle detection, and peer-to-peer communication to provide real-time collision warnings without any external hardware such as LiDAR or radar.
This project is built for Smart India Hackathon (SIH) 2025, based on the problem statement from ISRO, Department of Space: “Use of measurements from mobile phones to provide safe autonomous navigation on roads.”
| Type | Link |
|---|---|
| PowerPoint Presentation | Link 1 |
| App UI/UX Design | Link 2 |
| Full Documentation | Link 3 |
| Demo Video | Link 4 |
Every day, India loses 474 lives to road accidents. Two-wheelers—nearly 45% of total vehicles—lack built-in safety systems. Radar/LiDAR-based systems cost lakhs and are limited to luxury vehicles.
DRISHTI bridges this gap by enabling ADAS-like safety on any smartphone through:
- Raw GNSS + NavIC measurements
- IMU-based motion sensing
- AI-driven camera perception
- Peer-to-peer collaboration between vehicles
- Real-time collision prediction
The result: Affordable, scalable, India-first road safety technology.
- Uses GPS/NavIC, IMU, Camera + AI, and shared peer data
- Predictive algorithms estimate relative velocity and time-to-collision
- Alerts escalate: Safe → Warning → Danger
- Vehicles share state vectors using Wi-Fi Direct (preferred) or Firebase/WebSockets
- No internet required for P2P
- Multi-vehicle awareness improves accuracy and safety
- YOLOv8 Nano model via TensorFlow Lite
- Detects vehicles, pedestrians, static obstacles
- Works on low-end and mid-range smartphones
- Uses static smartphones as base stations
- Wi-Fi RTT + IMU for positioning
- Works in parking areas, warehouses, campuses
- Alerts for blind turns, intersections, accident-prone zones
- Uses Google Maps / OpenStreetMap layers
- Supports offline hazard mapping
- Analyses braking, acceleration, overspeeding
- Helps drivers improve habits
| Module | Description |
|---|---|
| Raw Data Acquisition | GNSS (GPS + NavIC), IMU, Camera |
| Sensor Fusion Engine | EKF fusion of GNSS + IMU + Vision |
| Cooperative Positioning | Wi-Fi Direct / Firebase P2P sharing |
| AI Object Detection | YOLOv8 Nano (mobile optimized) |
| Collision Prediction | TTC, trajectory modeling |
| Indoor Navigation | Wi-Fi RTT + IMU |
| UI/UX Layer | Map view + alerts |
| Cloud Optional | Firebase for sync, logs, fleet use |
Shared GNSS raw data allows:
- Common-view satellite differencing
- Reduction of clock/satellite errors
- Better relative accuracy than standalone GPS
For two vehicles:
[ t^* = -\frac{(r \cdot v)}{(v \cdot v)} ]
[ s^* = |r + vt^*| ]
Alert conditions:
- (s^* < \text{safe_distance})
- (0 < t^* < \text{time_threshold})
- Wi-Fi Round-Trip-Time (RTT)
- IMU dead-reckoning
- Anchors = static phones with known locations
- GPS + ISRO NavIC
- Accelerometer
- Gyroscope
- Magnetometer
- Camera (optional)
- Microphone (optional)
- YOLOv8 Nano (TFLite)
- Frame preprocessing (OpenCV)
- On-device inference (low latency)
- Wi-Fi Direct (P2P)
- Bluetooth LE (discovery)
- Firebase Realtime Database (fallback)
- WebSockets
- Android Studio (Kotlin/Java)
- Jetpack Compose / XML UI
- Google Maps SDK
- OpenStreetMap (optional)
- Firebase
- Node.js + Express
- Firestore/MongoDB
| Metric | Target |
|---|---|
| Outdoor Relative Accuracy | 1–3 meters |
| Indoor Accuracy | < 1 meter |
| Collision Alert Latency | < 200–500 ms |
| AI Obstacle Detection FPS | 15–20 FPS |
| End-to-End Delay | < 150 ms |
| Battery Consumption | < 20% per hour |
172 Indian Space Research Organisation (ISRO) Use of measurements from the mobile phones (low cost preferred) to provide a safe autonomous navigation on the roads Software SIH25177 Smart Vehicles
- India-first platform built for NavIC
- Fully decentralized P2P system (no server needed)
- No LiDAR, no radar — 100% smartphone-based
- Scales to millions of vehicles
- Works in rural areas, no internet required
- Complements Digital India, Smart Mobility, Atmanirbhar Bharat
- Core sensor fusion
- Wi-Fi Direct communication
- Basic collision alerts
- Static phone anchors
- Wi-Fi RTT positioning
- YOLOv8 Nano real-time detection
- Lane & signal detection (future)
- Firebase sync
- Fleet dashboard
- Driving behavior analytics
- Smart City APIs
- Insurance rewards for safe driving
- On-device AI (no video uploaded)
- Sensor data anonymized
- Encrypted communication (TLS/SSL)
- User consent required for all data access
git clone https://github.com/mrrogueknight/drishti.git
cd drishti
open in Android Studio
Build → Run on Android Device| S.No | Name | Role | Gender | Email ID | Mobile No. | Stream | Academic Year |
|---|---|---|---|---|---|---|---|
| 1 | Prashant Ranjan | Team Leader | M | 24MC3035@rgipt.ac.in | 8829013865 | B. Tech | 2nd Year |
| 2 | Ritik Prajapati | Team Member | M | 24mc3040@rgipt.ac.in | 7307255940 | B. Tech | 2nd Year |
| 3 | Ayush Pratap Singh | Team Member | M | 23ce3010@rgipt.ac.in | 7000267227 | B. Tech | 3rd Year |
| 4 | Sudeeksha Tripathi | Team Member | F | 24mc3050@rgipt.ac.in | 8899094625 | B. Tech | 2nd Year |
| 5 | Prateek Pandey | Team Member | M | 24mc3036@rgipt.ac.in | 9279721870 | B. Tech | 2nd Year |
| 6 | Karan Sharma | Team Member | M | 25cd3014@rgipt.ac.in | 9627379234 | B. Tech | 1st Year |
We thank the following contributors for their valuable efforts, collaboration, and contributions to this project:
- Prashant Ranjan – @MrRogueKnight
- Ritik Prajapati – @Ritik7307
- Ayush Pratap Singh – @Ayush-Pratap-Singh2006
- Sudeeksha Tripathi – @Sudeekshatripathi123
- Prateek Pandey – @Prateek1976
- Karan Sharma – @karansharmaworkspace
Choose based on your preference: MIT / Apache 2.0 / GPLv3
If you like this initiative, please ⭐ the repo and share it. Together we can build safer roads for India.