Advanced Adaptive Autopilot for Precision Guided Munitions

The opportunity exists to apply evolving technology for nonlinear adaptive control to autopilot design for precision guided munitions. The approach employs artificial neural networks to represent errors made in inverting the nonlinearities in a feedback inverting controller. Using an on-line method for weight adaptation results in a highly effective local approximator to the inversion error. Results indicate that highly accurate adaptation is achievable in fractions of a second, with minimal additional control effort employed during an adaptation interval. It is proposed that this methodology be employed in Phase I to develop an autopilot for a representative precision guided munition, and that the range of variation in munition design parameters be established over which adaptation is practical. It is anticipated that a single nonlinear adaptive autopilot could be developed that could be applied to a wide range of guided munitions and missiles, and that it would offer potential for dramatic improvement in weapon maneuverability via nonlinear adaptive control, especially in flight conditions were aerodynamics are poorly modeled, and for advanced configurations employing multiple control effectors with a very powerful resulting maneuvering capability.