Contains all simulink models and automated analysis used for the master's thesis Autonomous control of unmanned aerial multi-agent networks in confined spaces [Månsson, P., Green, S., 2019]. The project is initiated by running init_swarm_project.m. Complete and reusable models can be found in the lib directory, and working examples of those models can be found in controllers/outer_controller.
In this thesis, the functionality of existing Crazyflie models is extended to include current implementations of attitude and velocity control. The models are validated both as individual agents and as a swarm through comparisons of swarm behaviour during step, ramp and frequency response testing. The models are found to accurately replicate the dynamics of the real system, but additional research on measurement noise and accuracy would be necessary to ensure the model remains accurate in less than ideal conditions. Several suggestions for further improvements are presented. Additionally, two separate swarm controllers are implemented and tested in a simulated environment as well as the real world to demonstrate the capabilities of the model and to evaluate the controller usability in a number of practical use cases. Both controllers are found to behave well on the Crazyflie system, and demonstrate the practicality of the Crazyflie platform as well as the controllers implemented.
The contents of the folder models/greiff-2017 was developed by Marcus Greiff and is available through his mgreiff/crazyflie-project repo.