Kalman Filter as Observer and Smoother for Rigid-Body Motion Control Applications

Abstract

This paper addresses the problem of estimating non measured quantities in rigid body motion control applications resorting to Kalman filter theory. These quantities include the linear velocity of the rigid body, and external force and torque disturbances applied to its center of mass. The set of sensor measurements (pose and angular velocity readings) is noise-filtered as well. The overall framework consists of a linear time-varying system, shown to be uniformly completely observable, and an exponentially stable nonlinear controller. Simulation results are presented to validate and showcase the attainable performance of the proposed methodology.