Drones tracking based on robust Cubature Kalman-TBD-Multi-Bernoulli filter. (August 2021)

• Record Type: Journal Article
• Authors: Barbary, Mohamed; ElAzeem, Mohamed H. Abd
• Is Part Of: ISA transactions. Volume 114(2021)
• Page Start: 277
• Page End: 290
• Publication Date: 2021-08
• Subjects: Cubature Kalman Multi-Bernoulli Filter -- Micro-drones tracking -- Approximated variational Bayesian -- TBD; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2020.12.042
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: The problem of nonlinear tracking and detection of small unmanned aerial vehicles and micro-drone targets is very challenging and has received great attention recently. Recently, the Cubature Kalman-multi-Bernoulli filter which employs a third-degree spherical-radical cubature rule has been presented to handle the nonlinear models. The Cubature Kalman filter is more principled and accurate in mathematical terms. In addition, a recent multi-Bernoulli filter based on variational Bayesian approximation has been presented with the estimation the fluctuation of variances of measurement. However, Cubature Kalman and variational Bayesian-Multi-Bernoulli filters are unsuitable for tracking the micro-drones because of the unknown probability of detection. As we known, the track-before-detect (TBD) schemes was an effective method for tracking the small objects. In this work, a novel robust Cubature Kalman-Multi-Bernoulli filter with variational Bayesian-TBD is proposed jointing with estimate the fluctuated variances of measurement. The improved filter is an effective method to solve the problem of detection profile estimation for micro-drones. A novel implementation with a non-linear Cubature Kalman Gaussian mixture and Inverse Gamma approximation is presented to estimate a hybrid kinematic state of micro-drones. The simulation results confirm the effectiveness and robustness of the proposed algorithm. Highlights: Non-linear tracking schemes for small UAVs and drones from Image observations. A new robust Cubature Kalman-multi-Bernoulli (CK-MB) filter with variational Bayesian-Track-before-detect (VB-TBD) is proposed to cope with unknown measurement variances and Detection probability. New implementation include a non-linear cubature Kalman Gaussian inverse Gamma mixtures (CK-GIGM) solution, which applied to estimate augmented kinematic drones state under the same circumstance. The results show that proposed CK-VB-MB-TBD filter is more accurate than other kinds of Multi-Bernoulli filters in cardinality estimation and less Optimal Subpattern Assignment (OSPA) errors of small UAVs and drones tracking.