Project of drone swarm.
The purpose of this project is to test several algorithms, strategies and tools in the idea of creating a swarm of drones.
The project also have a self-training goal.
The purpose of the vox_drone module is to test the posibilities of using voxels instead of point cloud to represent the environment. The voxels indeed allow to use Fast Fourrier Transform and apply a low pass filter on the environment. Thanks to that, we can optains a representation of the environement with a very small memory footprint. That means we could possibly make the exange of the map between the drones faster or with a lower latency/communication frequency.
The future perspectives of this module are :
- Using a real map instead of a randomly generated one.
- test the computation time of the FFT and the compression.
- test of doing the compression and decompression faster, per exemple with AI.
- quantitatives test the impact of the compression on the quality of the map (Error rate, precision, etc.)
- precommit, hooks, documentation and refacto in the idea of making this module usable by other modules.
- data size optimization by using bit operations
source .config/setup.shpython3 src/main.py
The final result depend of the compression. The compression is currently parameted by a compression ratio : filter_size = map_size/compression_ratio
The purpose of the zawardo_drone module is to create a gazebo simulation of drone fleet that fit my need to test the other modules in a near use case simulation.
It is based on https://github.com/monemati/multiuav-gazebo-simulation.
Open3D Library : https://www.open3d.org/docs/release/index.html
Gazebo simulation for zawardo : https://github.com/monemati/multiuav-gazebo-simulation