Safe multi-agent drone control using control barrier functions and acceleration fields

The objective of this article is to develop real-time controllers for positioning multi-agent autonomous platforms (specifically drones) without any collisions. The framework is divided in two controllers. The first, based on control barrier functions and quadratic programming, is an acceleration field, that computes an acceleration for each position and velocity of the system that guides the agents towards the goal while avoiding obstacles and inter-collisions. The second controls the agents' total thrusts and angular velocities to track the previously -defined acceleration profiles. We show formal guarantees for both controllers, and simulation and experimental studies are used to show the efficacy of the proposed approach.