Approximate Feedback Linearization using Spline Functions

by Scott A. Bortoff


Abstract: This paper presents two numerical methods to synthesize an approximate feedback linearizing control for single-input nonlinear systems. Both techniques use part of the pseudolinearizing coordinate transformation as an ``output function,'' T1. The first synthesis uses a well-known recursive algorithm to compute the control given T1. Splines represent functions that cannot be computed in closed-form. The second method yields a simplified controller that can be computed regardless of the plant's complexity. Experimental results from a rotating inverted pendulum show approximate feedback linearization's improved performance compared with pseudolinearization, while highlighting the practical numerical difficulties that arise in the first method. Simulations of an acrobot illustrate the second method, which is also improved over pseudolinearization.


Return to S. A. Bortoff's publication list.