Drone Swarm Simulation

This application works to simulate drone swarm behavior in response to common group maneuver patterns. It uses a C++ backend to parallelize drone movement calculations across multiple threads, as well as a Java frontend for visualization.

The currently supported movement patterns are:

• Circling/hovering around a specified target within a certain minimum and maximum range
• Doing the above while maintaining a specified angular range of movement (for example, only circling between 0 degrees and 180 degrees when viewed from the top down)
• Returning to a central base location when battery is low for repair (battery consumption is calculated in Drone.cpp if you're curious about the exact calculations)

In addition, all movement is combined with obstacle avoidance and compensation for external wind speed. Currently, movement is calculated using a proportional controller, which works fairly well for most movement patterns. However, the steady state error for maneuvers tends to be a bit larger occasionally, so in the future a more comprehensive PI or alternative control model could be looked into (see World.cpp if you're curious about the details).

Using the Software

For reference, the C++ backend is executed by DroneSwarmSimulation.sln and drawing from the files in the \DroneSwarmSimulation folder, and the Java code is contained in the \DroneSwarmSimulation\DroneSwarmGUI folder.

Tasks are given by the Source.cpp file in the \DroneSwarmSimulation folder, simulating a central base station that assigns multiple tasks of varying priorities to the drone. In the nature of swarm robotics, each drone autonomously decides which task to execute based on its surrounding conditions, analyzing its distance to the target locations as well as the respective priority of each task to optimize how many drones are assigned to each task. This is done using a conventional cost function minimization approach (see World.cpp and Drone.cpp for the specifics).

Demo

A demo for the software can be found here: https://youtu.be/I5n_QFn6ilc. It shows several operations conducted by 100 drones on 4 targets, including multi-target tracking, obstacle avoidance, 360° surveillance, and returning to home.

Licenses and Acknowledgements

This work is licensed under the MIT License for future sharing and collaborative work. This software was developed during a 2017 internship at Elysian Labs, Inc. (https://www.elysianlabs.io/). It was done in collaboration with the AFSOC 720th OSS/OSK, with funding provided by the National Science Foundation (NSF).