A decentralized control of multiple unmanned aerial vehicles formation flight considering obstacle avoidance

Abstract

Fueled by the need to surpass the limitations of a single quadrotor, the formation control of multiple UAVs has become a research hotspot for many researchers. In this study, the authors present a decentralized behavior-based formation flight control that is capable to fly a group of quadrotors in an unknown environment avoiding obstacles and collisions between quadrotors. This objective is achieved by summing the weighted output of three behaviors, which are separation adjustment, trajectory tracking and obstacle avoidance, and sending the resulting control commands to the motors of each quadrotor. Using a cyclic directed graph topology, where each quadrotor makes use of the information of one of its neighbors, each connected pair of quadrotors maintain a specified separation. Collisions between quadrotors as well as obstacle avoidance are achieved using the artificial potential field approach. For the cascaded control of the quadrotor, integral-backstepping controller is used for the shape formation and PD controller is used for tracking the trajectory specified for each quadrotor. The decentralized formation flight strategy and the controllers are tested on a scenario of three quadrotors starting from different initial positions and required to fly in a horizontal line shape and to follow a ramp trajectory while avoiding collisions and obstacles. Results show the chosen strategy and controllers performed the desired task successfully.