Quadrotor Control based on Self-Tuning LQR

Recently, quadrotor has become one of the hotspots in robot research. To design an efficient controller for a quadrotor UAV is always challenging to for its under-actuated, strong coupling and nonlinearity. In this paper, a double closed-loop cascaded linear quadratic regulator (LQR) controller is proposed to track the referred position and attitude respectively. In order to solve the problem of tedious parameter setting, a parameter self-tuning control structure based on Simultaneous Perturbation Stochastic Approximation (SPSA) is designed and the parameters in the position and attitude controller are optimized respectively. By comparing with the original LQR controller, the simulation results show that the parameter-optimized controller improves the dynamic performance and control effect both in position control and attitude control, which verifies the correctness and feasibility of the proposed method.