Project Overview
Guardian is a GPS-denied autonomous firefighting drone system featuring vision-based localization and comprehensive mobile control. The system integrates a Raspberry Pi backend with Pixhawk flight controller, enabling autonomous navigation in GPS-challenged environments such as indoor facilities, dense forests, and urban canyons where traditional GPS-based drones fail.
The project features a Flutter-based mobile application for real-time drone control and monitoring, multi-client TCP server architecture supporting both operator control and external fire detection sensors, and advanced computer vision algorithms (SIFT + Optical Flow) for precise positioning. This comprehensive solution enables rapid emergency response with 60% faster deployment times compared to traditional methods.
Technical Architecture
Hardware Components
- Pixhawk 2.4.8 Flight Controller
- Raspberry Pi 4B (Onboard Computer)
- Camera Module for Vision Localization
- TFmini S Laser Rangefinder (Altitude)
- ESP32 Wi-Fi Hotspot Module
- Multi-rotor Quadcopter Frame
- Lithium-polymer Battery System
- MAVLink Serial Communication
Software & AI
- Flutter Mobile App (Android & iOS)
- Python TCP Multi-Client Server
- DroneKit-Python for MAVLink
- Computer Vision (SIFT + Optical Flow)
- Firebase Cloud Integration
- Real-time Telemetry Processing
- External Fire Detector Integration
- Background Services & Notifications
AI Fire Detection Model
Vision-Based Localization System
Guardian utilizes advanced computer vision algorithms for GPS-denied navigation. The system employs SIFT (Scale-Invariant Feature Transform) combined with Optical Flow to generate precise Vision Position Estimates (VPE) that feed directly into the Pixhawk flight controller, enabling autonomous operation in GPS-challenged environments.
Computer Vision Pipeline
- • SIFT feature detection and matching
- • Optical Flow motion tracking
- • Real-time pose estimation
- • Vision Position Estimates to Pixhawk
- • Raspberry Pi 4B processing unit
Fire Detection Integration
- • External fire detector sensor network
- • TCP-based alert messaging system
- • Real-time notification to operators
- • Sensor location mapping and routing
- • Automated mission coordination
Key Features
GPS-Denied Navigation
Vision-based localization using SIFT and Optical Flow for autonomous flight in GPS-challenged environments.
Fire Alert System
External fire detector integration with real-time TCP-based alerts and coordinate-based mission dispatch.
Mobile Control App
Cross-platform Flutter application for real-time telemetry, mission control, and drone status monitoring.
Multi-Client Architecture
TCP server supporting simultaneous admin and client connections with role-based access control.
MAVLink Integration
Direct Pixhawk flight controller integration via DroneKit for real hardware arming and mission execution.
Background Operations
Persistent TCP connection with background services for continuous monitoring and instant alert delivery.
Impact & Results
Project Achievements
Guardian successfully demonstrates GPS-denied autonomous navigation through vision-based localization, achieving stable flight in indoor and GPS-challenged environments. The integrated mobile application provides seamless operator control with real-time telemetry and fire alert management through a robust multi-client TCP architecture.
- Vision-based indoor navigation achieved
- Cross-platform mobile control deployed
- Real hardware flight control integration
- External sensor network compatibility
Technical Achievements
- ✓ SIFT + Optical Flow localization
- ✓ Pixhawk MAVLink integration
- ✓ Flutter cross-platform app
- ✓ Python multi-client TCP server
- ✓ Firebase cloud synchronization
System Capabilities
- → Autonomous mission execution
- → Real-time telemetry monitoring
- → Coordinate-based navigation
- → Role-based access control
- → Background service operations