Simulation 2024 05 / 14
Autonomous Drone Navigation
Sensor-driven drones that map, avoid obstacles and fly home — in 2D and then 3D.
Built with
Languages
- Python
- Java
Skills & tools
- Pygame
- Path planning
- Sensor modelling
- Autonomous navigation
Navigation algorithms for a simulated drone with distance sensors — obstacle-risk estimation, autonomous exploration and return-home — first in a 2D simulator, then extended to a full 3D environment.
From 2D to 3D
The first version moved a drone on a plane with simple sensor models and platform search. Going 3D meant rethinking almost everything:
- Movement in a full volume, not a plane
- Sensors that see above and below, not just around
- Autonomy that weighs vertical risk as well as horizontal
- Visualisation that makes the drone’s view understandable
How it decides
The drone casts rays to simulate its sensors, scores the risk of each direction — including up and down — and chooses a safe heading. It explores autonomously and returns home to recharge.
Gallery
-
The 3D simulator — the drone explores a map while its sensors report distances. -
Sensor rays sweeping for obstacles in the 3D view. -
“Where am I” — localisation on the maze maps.