Formation control of drone swarms via leader-follower maneuver regulation

The key idea of this thesis is to apply advanced formation control methods to swarms of drones, considering the leaders driven by maneuver regulation techniques. The approaches presented in this work propose an efficient solution to several cooperative multi-agent problems, e.g., in surveillance applications. In this thesis, the general dynamic model of the quadrotor is presented, paying particular attention to its differential flatness property. This property is exploited in the control technique of the quadrotor, where the centroid position vector and the yaw angle are taken as flat output vectors. Subsequently, the concept of maneuver is introduced, focusing on a circular path and proposing a maneuver regulation technique. This technique is widely applied in collision avoidance problems, given that the robot has to follow geometrical references. The adopted spatial trajectory is defined in terms of the circumference radius and tangential velocity. At this point, quadrotors are treated as general agents, belonging to a distributed cooperative network of agents, divided into leaders and followers. Indeed, a novel bearing-based approach is exploited to steer the followers to pursue a target formation, leaving the leaders unconstrained. For that reason, circular maneuver regulation techniques have been applied to them. The main contribution of this thesis is the validation of both the maneuver regulator and the bearing-based formation control technique. The simulation is based on the ROS 2 Toolbox ChoiRbot, where a cooperative multi-agent network can be easily managed. The code was implemented in Python, and thanks to its scalability with respect to the number of agents, formation algorithms with geometrical constraints of the maneuver can be performed.