ATGS

Air To Ground Scanner. Software to deploy on a drone for tracking targets using FLIR.

• Difference to the onboard software is that like a F-35 system you can connect different drone feeds to map out a field using machine learning. Smaller and faster drones can get closer to a target while flying on autopilot with commands from the homebase. This is to allow a command like receiver to have more situational awareness continuously. Meanwhile the tracking fleet can serve as overwatch and track human targets with FLIR at night and day.

• Mapping out targets using python ai for unmanned drone fleets. Really the goal here is to make a open source program and make it free to use, a poor man's version of DJI flighthub 2. Not only looking at cost but also make it available to use on none industrial drones like Mavic 3 or the mini series.

Hardware

• DJI Matrice 300 RTK Commercial Drone System

• FLIR Vue TZ20-R https://advexure.com/products/flir-vue-tz20-dual-thermal-camera?variant=35654698467483

Here's how this system works:

"ATGS" Launches a friendly user interface to control the server and drone.

"Drone" The drone feed pre-processes the captured video then sends it to the server. Using cv2 stream the video from the drone to the server and ultralytics Yolov8 for detection and annotation models.

"Server" This is the homebase to where different drones stream their video feed. Here is where the video window clients are launched.

Install

# Create python virtual environment
python3 -m venv venv

# Install dependencies
pip install -r requirements.txt

# Sometimes lap install doesn't work
pip install lapx

Execute

# launch 
python3 ATGS.py

Current State

• Operational

Disclaimer

This code for educational purposes only, do not attempt. I am not liable for any damages or injuries.