I've been interested in using technology for humanitarian purposes for quite a few years now. At the beginning of this hackathon, I did some research on search and rescue operations when people go missing; whether it be due to an uncharted hiking trip that has gone terribly wrong, or something as simple as sufferers of dementia wandering out of sight. Perhaps even after natural disasters when normal infrastructure falls short and hundreds of thousands are declared missing. After some more research, we found that a search party costs anywhere between $2000-7000 per hour and cannot get to certain places where the rescue workers' lives are put in jeopardy. Seeing how this is a problem, we sought out a solution.
What it does
Behold, the Manhunt-o-copter: capable of detecting lost souls in need of saving.
On a more serious note, we have developed a platform for use on existing unmanned aerial vehicles, or drones, as they are more commonly known, that can detect lost people on the ground, drop an aid package, and send a text message to a ground station with the GPS coordinates of the location where the lost person was found. In addition, a link is generated to a pin location on Google Maps for a one-click rescue operation.
How we built it
We used a existing RC helicopter as a representative model of our device, and we mounted our device around the frame. A webcam is mounted on the front, sending a video stream to the Jetson TX1, which ran a pretrained neural network to detect faces. Once our device detects a face, it is passed to a filter to remove false positives, and then passed into another script which activates a payload release mechanism to drop an aid package, in addition to receiving the GPS location from the flight computer and passing that to a ground station via SMS.