An autonomous quadrotor-mounted sensor system designed to maintain persistent visual surveillance of moving targets. Features custom mechanical design, real-time tracking algorithms, and synchronized multi-sensor payload integration.

Project Overview

POV (Persistence of Vision) is an autonomous quadrotor system designed to maintain continuous visual surveillance of moving targets. The system combines mechanical precision, real-time computer vision algorithms, and coordinated sensor control to track and monitor objects of interest with minimal human intervention.

This project integrates:

• Aerodynamic platform: Quadrotor-based aerial system for mobility and maneuverability
• Gimbal mechanism: Mechanically stabilized camera mount for steady visual tracking
• Vision processing: Real-time target detection and tracking algorithms
• Autonomous control: Flight controller integration with computer vision feedback

System Architecture

Overall Design

The POV system operates as an integrated platform combining:

1. Aerial Platform: Quadrotor with flight stabilization
2. Sensor Payload: Mounted camera system with real-time processing capability
3. Gimbal Control: Mechanical system for camera pointing and stabilization
4. Target Tracking: Vision-based tracking algorithms for autonomous surveillance

Hardware Architecture

Quadrotor Frame Design

• Configuration: X-frame layout with 4 rotors
• Motor system: Brushless DC motors with electronic speed controllers
• Battery: LiPo power source with voltage regulation
• Flight controller: Autopilot with IMU and barometer

Sensor Mount & Gimbal

Mechanical Design (CAD visualization in gallery):

• Camera bracket: Lightweight precision mount for stable mounting
• Gimbal mechanism: Single or dual-axis stabilization system
• Vibration isolation: Isolation mounts to minimize airframe vibration
• Integration points: Tool-less assembly for field maintenance

Features:

• Lightweight carbon fiber/aluminum construction
• Adjustable viewing angles
• Quick-release payload system
• Balanced mass distribution for flight stability

Sensor Payload

• Primary camera: High-speed camera for real-time tracking
• Processing unit: Onboard computer for vision algorithm execution
• Communication: Wireless link to ground control station
• Power management: Regulated power distribution for sensors

Software Architecture

Vision Tracking Pipeline

Camera Frame (1080p @ 30fps)
    ↓
[Target Detection - Object Recognition]
    ↓
[Centroid Calculation]
    ↓
[Motion Prediction]
    ↓
[Gimbal Orientation Command]
    ↓
[Flight Controller Adjustment]

Target Tracking Algorithm

The system processes video frames in real-time to:

1. Detect targets using computer vision techniques
2. Calculate target centroid in image space
3. Predict motion based on velocity vectors
4. Generate gimbal commands to keep target centered
5. Adjust quadrotor flight if target moves beyond gimbal range

Autonomous Control Modes

1. Gimbal tracking: Camera follows target while quadrotor maintains altitude
2. Full pursuit: Quadrotor follows target with gimbal maintaining lock
3. Surveillance: Circular pattern while tracking vertically
4. Return to home: Autonomous return to launch point when battery low

Key Features

Mechanical Design

• Precision machining: CNC-fabricated components for repeatable assembly
• Modular design: Quick-swap camera and gimbal systems
• Vibration damping: Isolated sensor mounts for image stability
• Lightweight structure: Optimized for quadrotor weight budget

Vision Capabilities

• Real-time tracking: Target detection and centroid calculation
• Motion compensation: Automatic gimbal adjustment for moving targets
• Autonomous operation: Minimal ground station intervention required
• Multi-target support: Ability to switch between targets

Operational Specifications

| Specification | Value | |—|—| | Flight time | 15-25 minutes | | Max speed | 12-15 m/s | | Altitude range | 0-120m AGL | | Camera resolution | 1080p @ 30fps | | Tracking accuracy | ±5° in gimbal range | | Communication range | 500m+ line-of-sight |

Technical Challenges & Solutions

Challenge 1: Gimbal Stability Under Motion

Solution: Gyroscopic stabilization with accelerometer feedback

• Real-time IMU data fusion
• PID control for smooth gimbal response
• Vibration isolation mounts

Challenge 2: Real-time Vision Processing

Solution: Optimized OpenCV algorithms on embedded hardware

• Multi-threaded processing pipeline
• GPU acceleration where available
• Adaptive tracking thresholds

Challenge 3: Flight Controller Integration

Solution: MAVLink protocol communication with autopilot

• Standardized communication interface
• Telemetry data logging
• Fail-safe Return-to-Home

Project Documentation

Complete project details are available in:

• Team Report PDF: Comprehensive design documentation, test results, and performance analysis
• Visual Assets: CAD design files, assembly photos, and system demonstration video

See the gallery above for all documentation and visual materials.

Conclusion

The POV system demonstrates the integration of mechanical design, embedded systems, and computer vision for autonomous surveillance applications. The modular architecture allows for adaptation to various missions, from research monitoring to industrial inspection.

Key achievements:

• ✅ Autonomous target tracking system fully operational
• ✅ Real-time gimbal stabilization with vision feedback
• ✅ Flight times exceeding design specifications
• ✅ Reliable autonomous operation in controlled environments

For more information and system demonstration, see the project report and video in the gallery above.