Inspiration

According to the WHO, Drowning is one of the ten leading causes of death for children in every region of the world. Of these, 26% of all unintentional drowning deaths are caused by river drowning. From personal family lost in river drownings, solving this issue holds a special place in the hearts of the team members

What it does

Afloat provides two products-

Scanner Drone - Deployed by a police station in the district next to the draining reservoir, the drone flies to the center of the lake and begins to take pictures periodically as it spirals outwards till the end of the lake. Once the drone comes back, it uploads the picture automatically to the ML engine

ML engine-The pictures taken from the drone are fed into a ML engine that picks out the anomaly pictures from the group and sends it to the police officer along with the GPS coordinates once it crosses the probability threshold. The officer takes the anomalous picture and decides whether to concentrate search and rescue responses in that spot instead.

How we built it

We built a custom drone from its basic parts from HobbyKing. We customized it to use inputs from a Arduino Uno to make it autonomous and controlled via a GPS module.We then controlled the Arduino Uno using a Raspberry Pi that would take pictures at the right moments and upload them to our ML engine. The extensive python development enabled us to then send an email out to the police officer in charge whenever the probability crossed a certain threshold.

Challenges we ran into

A lot of challenges popped up because of the integration between the 7 different parts of the project. Some of the major challenges faced was that one of the 4 cells that charged the custom drone gave away. We had a lot of issues integration with Microsoft's Custom Vision AI. However in the end, all the challenges that we faced taught us to perform and plan better for the real world.

Accomplishments that we're proud of

We are proud of our prototype because of the extent of the integrations. Also we were able to integrate the multiwii flight controller with the raspberry pi, enabling the use of the extensive python libraries that can integrate to the rest of the Internet of Things world.

What we learned

  • The power of Python
  • The ability to keep trying and to succeed, even after failing a million times
  • The nuances of electronic drones, batteries, flight controllers
  • The mystic world of Machine Learning

What's next for Afloat

As far as implementation is concerned, the next steps would be to

  • Test the product more extensively
  • Talk to first responders and see how we can tweak our product to make it more useful
  • Mass production and Open Sourcing

Special Thanks to

  • DJI Denver
  • Galvanize Boulder
  • AUVSI
  • HackDFW
  • XBuild18!!

Built With

Share this project:
×

Updates