IMPROVING MOBILE ROBOT LOCALIZATION USING EXTENDED KALMAN FILTER AND FUZZY LOGIC IN THE PRESENCE OF COLORED NOISE

In the present paper, a combination of Extended Kalman Filter and fuzzy logic has been proposed to optimally localize mobile robot in the presence of colored noise in system equations and observations vector. Observation vector is formed through having kinematic equations of mobile robots. For simultaneous positioning of mobile robot, it is firstly assumed that state and observations equations are contaminated with white uncorrelated noise with the zero mean and certain variance. Then, assuming that these white noises have passed through second-order filters, mobile robot localization problem is formulated through considering the resulted color noises. To this end, Extended Kalman Filter has been used to propose localization problem. To implement the mentioned algorithm, process noise covariance matrix can be estimated using various methods such as Taylor Series which has been employed in this research. On one hand, supposing that there is not so much information regarding process noise covariance matrix, fuzzy logic has been presented to estimate process noise covariance. Based on information received from innovation vector, fuzzy logic performs the simultaneous updating of the variance. The results obtained from simulation have been presented in low, average and high variances and the proposed localization performance have been investigated based on estimation accuracy

Keywords:

Localization , Mobile robot , Extended Kalman filter , Fuzzy logic , Colored noise