Non-linear adaptive auto-pilot for uninhabited aerial combat vehicles Conference Paper uri icon


  • © 1999 by M. R. Akella and I & mesh S. Published by the American Institute of Aeronautics and Astronautics, Inc. We demonstrate global stability and bounded tracking for a class of model reference adaptive controllers (MRAC), in the presence of bounded disturbances and model errors. The MRAC problem is cast into the Structured MRAC problem by imposing the condition that the kinematic differential equations are exactly known, and all the learning is restricted to the acceleration differential equations. Earlier work in this area considered the number of actuators to be equal to or greater than the number of velocity states. A class of problems wherein, the number of actuators are less than the velocity states is addressed here and the adaptive control laws are designed to ensure global stability and bounded tracking. It can be proved in theory that, the present case could be cast into a structure that seeks to drive the controls by minimizing the tracking errors in a least squares sense. The controller structure is designed, seeking to drive the error energy rate as negative as possible, and we show that the closed loop system is globally stable. However no claim is made with regards to convergence of the adaptation rates and parameters. The overall performance of the system is limited to bounded tracking, i. e., allowing the errors to remain within desirable bounds. The above methodology was applied to track a very aggressive maneuver of a high performance aircraft which was a transition from a wings level, straight line flight to a bank of 60° and subsequently a heading change of 180° maintaining constant speed and at altitude of only 1000 ft. An interesting feature of the whole exercise was the refinement of the mathematical model of the plant by incorporating the structured kinematic nonlinearities arising out of gyroscopic coupling and coriolis terms thereby rendering the plant to have "almost" exact nonlinear structure. This presents us with a model, which is now uncertain only in it's inertias, aerodynamics and the propulsive influences which is usually the case physically. The resulting structured adaptive control formulation is thus brought to a canonical form wherein there is a physical justification for the adaptive process. The control law is nonlinear in two respects, the nonlinear plant itself and the non-linear adaptation process.

author list (cited authors)

  • Akella, M. R., Subbarao, K., & Junkins, J. L.

publication date

  • January 1999