Eigenvalue Based Aircraft Autopilot Design is Non-Robust: New Robust Control Design Using Transformation Allergic Indices

In this paper (with its IP protected contents), we demonstrate that the current literature, eigenvalue based Autopilot Design (labeled as Transformation Compliant (TC) design) suffers from hidden, real state variable (flight dynamics) Divergence (i.e. Instability), not recognized by the TC methods. This instability is labeled in this paper as Blind Spot (BS), Non-Phase, Real (NPR) State Variable Divergence (SVD) (i.e. (BS-NPR-SVD). We then show that the analysis and control design proposed in this paper does not allow such instability to happen at all and proposes a new Autopilot design with Safety Assurance. The approach proposed in this paper is labeled as Transformation Allergic (TA) approach in contrast to the Transformation Compliant (TC) approach of the current literature. It is urged in this paper that Autopilot designers using currently prevailing (since almost a century) popular TC approaches which are based on the use of eigenvalues as state variable convergence measures need to take this Blind Spot, Non-Phase, Real State Variable Divergence very seriously. The proposed TA approach uses Transformation Allergic Indices (TAIs) as the real state variable convergence measures, doing away with eigenvalues completely. Illustrative examples with realistic aircraft data used in the current literature models are used to demonstrate this BS-NPR-SVD phenomenon. The proposed TA approach highlights the importance of the signs and magnitudes of the aircraft stability derivatives which are shown to play a critical role in achieving Convex Convergence of aircraft related physics based, real state variables rather than depending on the misleading conclusions drawn by the convergence properties of unrealistic, non-physics based, non-linearly transformed state variables, namely, phase variables (which are masquerading as linearly transformed state variables) that are being used in the current literature TC methods.