Fast automatic leveling subject to abrupt deterministic input

Abstract

This investigation applies a modified Kalman filter with a recursive generalized M estimator (GME) of input to a class of leveling problems, that are subject to abrupt environmental disturbances and high noise levels. A least-squares estimator (LSE) based hypothetical testing scheme is also devised to detect the onset and presence of the input. Simulation results demonstrate that the leveling speed of convergence and accuracy is markedly higher than the original unmodified one.