Kavach is an experimental autonomous UAV system focused on real-time perception, decision-making, and control using computer vision and PX4-based flight systems.
To design and test autonomous drone behaviors that integrate perception, navigation, and control logic in realistic simulation environments.
- Real-time object detection and tracking
- Autonomous navigation logic
- PX4-based flight control integration
- Simulation-first development approach
- YOLO (Object Detection)
- OpenCV
- Computer Vision
- PX4 Autopilot
- MAVSDK
- SITL / Gazebo Simulation
- Python
Advanced autonomous maneuver simulations for tactical scenarios.
| Vehicle Type | Model | Key Behaviors |
|---|---|---|
| Multicopter | typhoon_h480 |
High-speed ascent (6m/s), Hover stability, 35° Dive |
| Fixed-Wing | plane_cam |
Runway takeoff, Loiter orbit, High-speed Dive |
Key Features:
- Modular mission architecture
- Automated safety checks (Altitude floor, Tilt limits)
- Comprehensive telemetry logging (
logs/directory)
Kavach/
├── Diving_Scripts/ # Autonomous maneuver logic
│ ├── multicopter/ # Quadcopter-specific missions
│ ├── fixedwing/ # Plane-specific missions
│ ├── logs/ # Mission telemetry logs
│ └── legacy/ # Archived tests
├── models/ # Custom Gazebo models
└── README.md # Project documentation
- Improved tracking and prediction
- Advanced control strategies
- Transition from simulation to real hardware