Shuttle reentry guidance using Sugeno approximation

The Orbiter guidance law is required to track a reference state trajectory that satisfies a given drag-acceleration versus velocity profile designed to maintain the structural and thermal integrity of the Shuttle Orbiter. New nonlinear guidance laws have been proposed that require complicated multidimensional interpolation. In this paper, we use Sugeno approximators (hybrid fuzzy-crisp inference engines) to conduct such interpolation. The Sugeno approximators are trained by example using a recursive least-squares algorithm similar to a static Kalman filter. Moreover, we introduce the concept of surface-tracking guidance: If disturbances and sensor noises are severe enough to make the Orbiter deviate significantly from the desired reference trajectory, the law control authority excludes effective tracking unless the reference trajectory itself is updated. We attempt to implement an “online optimizer” using the interpolating and learning capability of Sugeno approximators. When applied to the Shuttle reentry problem, the “online optimizer” is a constellation of few blocks two of which are MIMO Sugeno approximators. We study and simulate the different approximators and guidance laws