Fuzzy Logic Augmented H Infinity Filter for Target Tracking

Abstract

The demand of precise and uncertainty capable robust target designs in real time has taken drastic growth in recent time. The performance of Conventional Kalman and H-infinity filter and its variants is affected in presence of outliers and behaves unstable for non-linear devices. Study presented here discusses adoption of modified H Infinity Filters for multi sensor data fusion based target tracking. Fuzzy Logic based H-Infinity Fuzzified Filter (FLHIFF) is presented where in initially fuzzy logic is used at filter level to eliminate local estimation errors and an additional fuzzy system is used to minimize outlier effects and estimation errors during data fusion. In addition, adoption of Adaptive H-infinity Filter (AHIF) for multi sensor data fusion MSDF based non-linear tracking is also presented. In AHIFFDoM Degree of Matching (DoM) is computed and in accordance to it process noise covariance matrix is tuned using fuzzy logic to achieve matching. Performance of FLHIFF is compared with classical H Infinity filters for mild and evasive maneuvering targets through experiments. Superior performance of FLHIFF for mild and evasive maneuvering target tracking is proved. Superior performance of AHIFFDoM when compared to state of art existing Kalman based existing technique for target tracking is proved based on experimental results obtained. newlineSimilarly, Adaptive H-Infinity Filter can work efficiently in the presence of uncertainties using sliding window concept. In our proposed AHIF, length of window size is varied to eliminate estimation errors and predict precise location of target. Experiments conducted to evaluate performance of AHIF in comparison with classical Kalman and H Infinity filters for mild and evasive maneuvering targets. Results verify the superior performance of AHIF for mild and evasive maneuvering target tracking. Our results demonstrate that AHIF performance in comparison to conventional Kalman and H-infinity filter is better in terms location accuracy and Position Fit Error (PFE). newlineA fuzzy logic and H-infinity based nonlinear observer for continuous time dynamic system is presented. Fuzzy logic membership function (FMF) operates on the observer residuals as a modulating artifice. The Lyapunov energy (LE) functional is used to derive condition for the local asymptotic stability for this H-infinity (HI) based observer s state errors. newline