Observer Design for Vector Nonlinear Systems With Applications to Autonomous Vehicles and Robots

ABSTRACT

This paper focuses on observer design for nonlinear systems involving functions of vector arguments. The output equations and the process dynamics are both composed of nonlinear functions of nonseparable multivariable states. The partial derivatives of the nonlinear functions are assumed to be bounded. A nonlinear observer with a single constant observer gain is developed and an observer gain design LMI for exponential stability is obtained. Furthermore, a scaling of states technique is also introduced to enhance numerical solutions for LMI feasibility. The observer developed in this paper is used for two applications in which previous scalar argument observers do not work. The first application is state estimation of the ego vehicle for both highway and local driving. The second application is roll and pitch angle estimation for legged robots. Experimental results are presented and verify the performance of the developed methodology.