A Probabilistically Robust Eigenstructure Assignment Technique for Flight Control Design of UAVs

Abstract

This paper proposes a technique that uses a randomized algorithm to assign robust eigen-structures for a multivariable system, such as Unmanned Aerial Vehicles (UAVs). The freedom offered by a multi-input state feedback controller, beyond a specification of distinct closed loop eigenvalue placements, is utilized to shape the transient response as well as the robust stability of the UAV. A mixed time and frequency domain performance index under parametric system uncertainties is considered while designing the robust controller based on eigen-structure assignment. Ensuring robust stability and performance with a guaranteed transient response against structured parametric uncertainties for a linear MIMO systems is in general NP-Hard and thus computationally intractable. To tackle the above-mentioned difficulties, an algorithm based on randomized methods is presented. The randomized algorithm guarantees good performance in the “average case” over all possible scenarios and does that in polynomial time. Moreover, it finds an approximate global minima of a joint performance index by a procedure of selective sampling and evaluating the design parameter space. Effectiveness of the algorithm is shown with extensive numerical simulations of different UAV flight modes, such as longitudinal and lateral modes.