Embedded Files
Inspiration
This project was my very first hackathon, marking the beginning of my passion for rapid hardware challenges. Taking place during the severe California wildfire season, my teammate and I were inspired to build a preventative solution: an automated, portable tracking turret designed to actively seek out and target wildfires for rapid extinguishing.
Fabrication steps
Computer Vision:
Computer Vision:
Overcame the lack of a thermal sensor by utilizing OpenCV to perform real-time color filtering. The system isolated bright orange color spaces (simulating fire) via a laptop webcam.
Targeting Logic:
Targeting Logic:
Developed an algorithm to dynamically draw bounding circles around detected objects, prioritizing the largest diameter to ensure the turret always tracked the closest and most immediate threat.
Resource Constraints
As a beginner at a hackathon, I had very limited access to specialized hardware. We didn't have external camera modules or expensive thermal imaging sensors, so we had to be highly resourceful. We decided to drive the system using an Arduino Uno and standard servo motors, relying on my laptop's built-in webcam as the primary visual sensor. I quickly modeled a custom 2-axis tracking mount in Fusion 360 and 3D printed the entire assembly on-site.
Tracking Algorithm
Without a heat sensor, I had to find a software workaround to detect "fire." I utilized OpenCV's image processing libraries to apply color filtering masks to the live video feed, isolating the bright orange wavelengths characteristic of a flame. To test and calibrate the system safely, we used a bright orange ping-pong ball as our target.
Once the software could "see" the orange object, I wrote an algorithm to calculate its size and position. The script drew a bounding circle around the detected color space and calculated its diameter. This ensured that if multiple "fires" were detected, the system would prioritize tracking the closest and largest threat. The software then mapped the object's X/Y pixel coordinates to physical angles, sending continuous movement instructions to the Arduino.
filters applied to feed
To accurately track the ping pong ball, we simply used geometry! we measured the distance from the ball and camera and since it formed a triangle, we could find the third side using pythagorean theorem
Of course since we were limited on resources, we were working the best we could and even got to modeling on fusion! I was able to 3D print everythhing and we surprisingly achieved some accurate results.
20250126_112546.mp4
To prove the physical accuracy of our software-to-hardware pipeline, we attached a laser pointer to the 3D-printed turret head. During the demonstration, the turret tracked the orange ping-pong ball with surprising precision. It was an incredibly successful first hackathon that provided me with foundational experience in CAD, microcontrollers, and computer vision.
Page updated
Google Sites
Report abuse