Application of SPSA LQR tuning on quadrotor

Abstract

Systematic LQR tuning is necessary to improve trial-and-error method that is usually used in LQR tuning. This paper presents LQR tuning with SPSA applied on quadrotor control. SPSA is chosen because it is a powerful algorithm for multivariate stochastic optimization and can be easily implemented. Numerical simulation shows that LQR tuning with SPSA can yield good quadrotor control performance. However, it becomes apparent that the performance is insensitive toward LQR parameter changes.