Open Source Search & Rescue Grid Mission Tool for Parrot Anafi: Introducing P6.Scan

Hello Parrot community,

I’m excited to share a project developed as part of our bachelor’s in Software Engineering at Aalborg University, which may be of interest to those still flying Parrot Anafi drones or exploring Parrot’s developer ecosystem.

Project Overview:
Our project, P6.Scan (“Project 6” for our 6th semester, “Scan” for self-controlled aerial navigation), is an open-source search and rescue tool designed for the Parrot Anafi. It leverages the drone’s SDK, documentation, and the Parrot Sphinx simulation environment to provide autonomous grid-based area scanning for locating people in need of rescue.

Unfortunately I am unable to link videos because I’m a new user, but there is a video demonstration on my youtube: (my username: @EmilPrivate) (video name on YT: Self Controlled Aerial Navigation (SCAN) - Aalborg University Bachelor’s Project - P6). I have also open sourced the entire project on my github: (my username: @Emilprivate) in my Esd.EDU repository.

How it Works:

  • The operator selects an area on a map using a web frontend.

  • The backend calculates an optimal coverage grid using a custom Coverage Path Planning (CPP) algorithm with TSP (Traveling Salesman Problem) optimization.

  • The drone autonomously flies the grid, takes photos, and a custom-trained YOLO computer vision model processes images to detect humans in water, but due to lack of resources, the model is very limited with its accuracy, in-fact an R-CNN model like ResNet would be more ideal for this kind of project.

  • Detected locations are mapped and displayed to the operator.

  • After the mission, the drone returns and lands automatically.

Why Share Here?
While the Anafi is no longer the newest platform, its SDK and simulation tools are still very interesting for prototyping and educational purposes. If you have an Anafi or are interested in adapting the concept for other Parrot drones, I hope this project can be a useful resource or inspiration.

If you have questions, feedback, or ideas for adapting the project, I’d love to hear from you!

2 Likes

Great project!

Just in case you were not aware, there is a world called parrot-ue4-planet that allows you to spawn the drone at real-world locations. All it requires is a free Google API key. It would ensure that the map view accurately reflects the simulated environment.

You can also add actors to the scene for detection purposes: Populate the scene with actors - 2.15.1 (though note that this feature is not available in the Planet world).

Thanks for the suggestions, Ocrave!

I wasn’t aware of the parrot-ue4-planet world - that sounds really useful for making simulations more realistic with real-world locations. The Google API integration for accurate map views would definitely improve our project.

I’ll look into both features you mentioned. The actor population could also help with testing our detection algorithms in different scenarios.

Appreciate you sharing these resources!