Fast nonlinear model predictive control for multicopter attitude tracking on SO(3)

摘要

Exploiting the highly dynamic flight envelope of a multirotor Micro Aerial Vehicle (MAV) is a particularly challenging task that requires special treatment of its attitude control loop. In this paper, we propose a fast nonlinear model predictive control approach based on a geometric formulation of the error to track the MAV attitude on the SO(3) special orthogonal group. The proposed controller is combined with an optimal position tracking control strategy in a cascaded fashion. The overall framework is implemented on-board a hexacopter and verified in both high fidelity simulations as well as extensive experimental studies. As shown, the resulting closed-loop system exploits the dynamics of the platform, is able to track aggressive trajectories, recover from arbitrary attitude configurations and also handle a propeller failure.