Sensor Fusion for simple walking robot using low-level implementation of Extended Kalman Filter

Abstract

The main aim of this paper depicts in design and implementation of the Extended Kalman Filter for a nonlinear system in an application of a sensor fusion from a practical point of view. The sensor fusion is a typical data processing problem in mechanical systems where individual measurements of (angular) positions, velocities or accelerations are done independently on each other but the measured values are correlated to each other via dynamics of the system. Moreover, the measurement is corrupted by noise. The sensor fusion technique is capable to gain proper information about positions, velocities or accelerations from inaccurate measurement. In background of the sensor fusion algorithm, in our particular case, works the Extended Kalman Filter. Its adaptation for a simple mechanical system represented by a nonlinear system are object of the research in this paper related to usage of the Extended Kalman Filter on a low cost hardware.