AutonomousRescueDrone

• 

  Initial versions of object detection within flight

Inspiration

Our team comes from Pakistan, ana active tectonic zone where severe earthquakes are common. Due to being a developing nation the lack of resources hinders the ability to save survivors in difficult situations. We hope to bring around a more convenient, time-saving solution.

What it does

The Drone is activated through a laptop. After taking off, it looks around for any faces, bodies it can detect. Once a person is detected, it centers onto their image autonomously.

How we built it

We used the Tello SDK to control the hardware. We then used OpenCV on the drone feed for object detection and finally programmed the drone to achieve semi-autonomous flight for this purpose in python.

Challenges we ran into

The drone has a very limited range, no proximity sensors, GPS so for the demo we had to write a program to make sure that the drone works in a controlled space as a proof of concept.

Accomplishments that we're proud of

We are proud of the fact that we have been able to achieve semi-autonomous flight in under less than 24 hrs and with a team that includes two newbie coders. None of us had any experience with computer vision in the past.

What we learned

We learnt the complexities of reliably controlling a drone with a UDP connection, using object detection with a live feed, being able to jerry-rig functions to randomize the autonomous search algorithm.

What's next for AutonomousRescueDrone

If we are able to get some funds to buy a drone with the necessary hardware to scale the project, we would like to create a full scale application for the project. We also plan to create a web-app to make it user friendly to control the application.

Built With

• opencv
• python
• tello