Real-time implementation of decentralized adaptive formation control on multi-quadrotor systems

In this study, we focus on real-time implementations of a practical distributed adaptive formation control scheme for a multi-quadrotor system with uncertain inertial system parameters. We design a decentralized controller based on the leader-follower formation approach to motion control of such a system in rigid formation. The proposed control approach has a two-level structure: At the high level, a distributed control scheme is designed for the kinematic formation control problem. In the low-level, we analyze each single quadrotor control design in three parts. The first is an adaptive linear quadratic controller under consideration of inertial uncertainties for the pitch and roll dynamics, and in this case, we design an on-line parameter estimation with the least squares approach, excepting yaw and altitude dynamics. The second is proportional controller for the yaw motion. The third is proportional-integral-derivative controller for altitude. Throughout the formation control implementations, the controllers successfully satisfy the formation maintenance objective. Simulations and experimental results are presented considering various scenarios, demonstrating the effectiveness of our algorithm.