Autonomous Drone Swarm: coordinated flight in simulation

What the experiment does

An engineering experiment in which several drones fly together inside a computer simulation. We use it to test how a group of machines moves in coordination without a pilot steering each one. The drones hold their formation, avoid collisions and carry out a planned mission. Systems that direct fleets of vehicles or devices need the same logic.

How it works

The experiment starts with a mission for the whole swarm. The planner turns the mission into commands for each drone and gives each one its place in the formation. The commands reach each drone's autopilot over a communication protocol designed for drones. Each drone sends back its position and status continuously. The system compares those readings with the target formation and corrects any drone that drifts out of place. When two drones come too close, it adjusts both paths before they meet. Every step runs inside a physics simulation, so a mistake shows up on screen before it costs anything. The result is a swarm that carries out its mission while holding formation.

What we can build for you

• We build systems that track many vehicles or devices in real time and send each one its next command.
• We test coordination logic in simulation before any field trial, so mistakes surface while they are cheaper to fix.
• We connect your system to hardware over open, documented protocols that any engineer after us can read and extend.

How it is built

ROS 2 links the parts of the software so they exchange messages in real time. MAVLink, a protocol common across drone autopilots, carries commands out and readings back. ArduPilot flies each drone; it is an open-source autopilot that also runs on real aircraft. Gazebo provides the 3D physics simulation the drones fly in, and it works directly with ROS 2 and ArduPilot.

• ROS 2
• MAVLink
• ArduPilot
• Gazebo