Autonomous UAV Flight and Remote Lighting Control System

【无人机演示】https://www.bilibili.com/video/BV1wDwQemEV3?vd_source=003215dcbf37b2c6060f4782a5c7afa5

Project Overview

This project implements a multi-functional UAV system that combines autonomous flight control, real-time lighting optimization, and adaptive posture adjustment. The system is designed to optimize low-altitude detection by dynamically adjusting the onboard lighting and the UAV’s posture based on real-time environmental conditions. The core technology is based on Continuous-Time Neural Networks (CTNN), which allows the UAV to continuously adapt to changes in lighting and flight posture, ensuring optimal detection performance.

---

Key Features

• Autonomous UAV Flight Control: The UAV autonomously navigates a predefined flight path using MAVROS for offboard control.
• Real-time Lighting Adjustment: The onboard lighting is dynamically adjusted based on the lighting conditions and real-time image analysis, powered by a CTNN-based model.
• Posture Adjustment: The UAV adjusts its posture (pitch, roll, yaw) in real-time to optimize sensor performance and enhance image quality during flight.
• Embedded System Deployment: The system is designed for deployment on Jetson Orin Nano, with real-time image processing and control.

---

Technical Approach

1. Autonomous Flight Control

• The UAV uses MAVROS for offboard control, integrating it with PX4 for flight state management and navigation.
• The system monitors UAV states, adjusts flight position, and controls the altitude with real-time updates from the onboard IMU and GPS.

2. CTNN-based Adaptive Lighting Optimization

• CTNN (Continuous-Time Neural Networks) is used to model the relationship between lighting intensity and image quality.
• The UAV adjusts the LED lighting intensity through PWM based on the output of the CTNN model.
• The model uses real-time image inputs and flight posture data to predict the optimal lighting intensity that improves target detection accuracy.

3. Real-time Posture Adjustment

• Posture optimization is handled by the same CTNN, which uses the UAV’s current posture (pitch, roll, yaw) and image quality as inputs.
• The system computes the best posture configuration to maximize the effectiveness of the lighting and sensor data collection, ensuring optimal target detection.

4. Real-time Image Processing and Control

• ZeroMQ is used for real-time image transmission from the UAV’s camera to the ground station.
• A remote control system, using XML-RPC, allows for dynamic adjustment of lighting parameters via a web-based interface.

---

System Components

• Jetson Orin Nano: Embedded platform for running the CTNN model, controlling PWM, and handling image processing.
• MAVROS & PX4: Software stack for flight control and navigation.
• CTNN Model: A neural network used to predict the optimal lighting intensity and UAV posture adjustments.
• LED Lighting: RGB LEDs for adjustable lighting, controlled via GPIO pins on the Jetson Orin Nano.
• ZeroMQ: Real-time communication protocol for transmitting images to the ground station.
• XML-RPC: For remote control of lighting adjustments via the ground station.

---

Installation

1. Prerequisites

Ensure you have the following installed:

• ROS Noetic (or a compatible version)
• MAVROS package
• Jetson Orin Nano setup with Ubuntu 20.04 and Jetson.GPIO
• ZeroMQ library for Python
• PyTorch for running the CTNN model

2. Clone the Repository

git clone https://github.com/yourusername/Autonomous-UAV-Control.git
cd Autonomous-UAV-Control

3. Install Dependencies

# Install ROS dependencies
sudo apt-get install ros-noetic-mavros ros-noetic-mavros-extras

# Install ZeroMQ
pip install pyzmq

# Install PyTorch
pip install torch torchvision

4. Set Up MAVROS

Ensure that MAVROS is configured correctly for communication with your UAV's flight controller (PX4).

roslaunch mavros px4.launch

5. Run the System

Start the UAV control and monitoring system:

# On the UAV machine (Jetson Orin Nano)
roslaunch autonomous_uav_control start_flight_control.launch

# On the ground station
python run_ground_station.py

---

Usage

• Flight Control: Once the system is running, the UAV will autonomously take off, follow the predefined flight path, and adjust its posture to optimize target detection.
• Lighting Adjustment: The UAV dynamically adjusts its onboard lighting based on the current environmental conditions, with the lighting intensity controlled by PWM.
• Real-time Monitoring: The system allows for real-time monitoring of the UAV’s flight state, posture, and lighting through a ground control interface.

---

Future Work

• Multi-source Lighting Optimization: Future enhancements could include RGB dynamic lighting control to adjust the color temperature for various environments.
• Enhanced Posture Adjustment: Integration with additional sensors like IMUs and LIDAR for more precise posture optimization.
• Real-time Feedback Loop: Implement a more advanced reinforcement learning approach for real-time feedback-based optimization of lighting and posture.

---

Contributors

• Your Name – Lead Developer
• Collaborator Name – UAV Flight Control Specialist
• Collaborator Name – Computer Vision and Model Optimization Specialist

---

License

This project is licensed under the MIT License – see the LICENSE file for details.

---

Feel free to adapt the README as necessary based on your project specifics!