Search and Rescue Smart Drone Project


I am currently working on a search and rescue drone for National Parks that uses machine learning to detect missing people from a 3DR Solo.  If you are interested, please check out our website, I would love to hear feedback from fellow drone enthusiasts!  ARI is an unmanned aerial system capable of detecting people from aerial video data during the initial search for a missing person.OUR MISSIONARI is designed to increase the chances of finding a missing person using the newest drone and machine learning technology. The system is modified to fit the methods of search and rescue teams of the National Parks so that it can be easily integrated without disrupting the current search processes.HOW IT WORKSAt the start of a search for a missing person, a drone is deployed and autonomously scans the area. RGB and thermal video from the flight is quickly processed through a trained neural network from Amazon Machine Learning to determine the exact video frame and GPS location at which a person was found. All data collected can be viewed on a GIS map, allowing members of the Search and Rescue team access to the locations of the found person.OUR PLATFORMOur platform is the 3DR Solo, which is already being used for search and rescue throughout the National Park System. ARI uses and RGB camera as well as FLIR thermal sensors for more accurate data and the ability to perform searches during the night.OUR COMPANYARI began as an undergraduate research project at the University of California, Merced, and is now competing in the Berkeley Big Ideas Competition for Spring 2017 in the Hardware for Good category.WHAT WE'VE ACHIEVEDInitial stages of training the neural networkArcGIS online capabilitiesSystem integration into existing SAR methodsOUR FUTURE GOALSGet full system working for Yosemite National Park to useSuccessfully help find missing personsUse for other applications such as wildlife research

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • Nikola...good point! If machine learning is being used, give it another layer of information to parse data, and the system will be more effective. This is some very interesting research. I sent you an email, from

    My web presence is currently on Facebook at:

    Please feel free to contact me about further discussion of this topic, as it is something in which I am personally interested, and am conducting research myself.
  • Talk to NASAR.  Also talk to Prof. David Lovelock at ASU.  You can optimize your search areas using lost person behavior statistics.

This reply was deleted.