Higher-order implicit Newton observer with application to trajectory tracking

Abstract

This work aims to extend the Newton observer design to time-varying nonlinear systems using higher-order implicit Runge-Kutta methods. The proposed observer can produce accurate state estimations at longer sampling times, where explicit methods typically fail. The Newton observer is further robustified by including disturbance estimation and implementing warmstarting techniques. A scheme to combine the implicit Newton observer with higher-order sampled-data control is presented to realize trajectory tracking at longer sampling times. Experiments on a magnetic levitation system is conducted as validation. The performance of the implementation with different orders is evaluated with respect to accuracy, robustness and computational cost.