Field-Testing of High-Level Decentralized Controllers for a Multi-Function Drone Swarm

Abstract

This paper describes the results of initial field testing of a high-level decentralized controller for a swarm of multi-rotor drones. This controller allows the intuitive implementation of a wide variety of swarm behaviors. The parameters of this controller were mapped to two possible missions. The proposed controller structure is parameterized to enable the evolution of swarm behaviors in future work. In this paper, a set of hand coded controllers, previously evaluated in simulator were field tested. The simulator used considers each UAV as a simple point-mass model of each agent. As such, the ability to transfer the controller structure from simulated UAVs to real UAVs is investigated. The experiments use a fleet of enhanced networking-capable commercial-of-the-shelf drones that were especially developed to support and demonstrate the proposed high-level controller in the decentralized swarm environment. The paper describes hardware and controller implementation and then proves feasibility and great potential of the proposed approach in a series of flight tests conducted by a swarm of four identical drones.