Search

In this paper, a novel maximum fuzzy correntropy Kalman filter (MFC-KF) algorithm is proposed to solve the problem that the effect of different samples on state estimation is uncertain in common correntropy. In the proposed algorithm, a new optimization criterion—the maximum fuzzy correntropy criterion with fuzzy correntropy based on fuzzy information theory—is used to optimize the Kalman filter, by reducing the effect of the common correntropy applying the same weight for all samples. Moreover, to apply the MFC-KF algorithm to bearings-only maneuvering target tracking, it is combined with the least-squares method for measurement conversion. Moreover, the kernel width is set adaptively. Simulations show that the proposed algorithm can track a target more accurately than the interactive multi-model extended Kalman filter (IMMEKF), the interactive multi-model unscented Kalman filter (IMMUKF), or the maximum correntropy Kalman filter (MCKF).

This paper proposes an interacting Takagi–Sugeno(T-S) fuzzy semantic model estimator (ITS-FSM) for maneuvering target tracking, and constructs a framework for a generic interacting T-S fuzzy semantic model to incorporate semantic information concerning the target. To adaptively calculate the transition probability matrix of the models, a probabilistic switching model based on semantic fuzzy sets is derived by using the degree of intersection between fuzzy sets, which enables switching from one semantic fuzzy set to another. We also propose an efficient kernel maximum entropy fuzzy clustering method to identify the premise parameters of the model. This enables the proposed algorithm to recursively estimate the premise parameters in case of a limited number of samples. Moreover, a modified extended forgetting factor recursive least squares (MEFRLS) estimator is used to identify the parameters of the T-S fuzzy semantic model. The results of experiments on three simulation datasets show that the proposed ITS-FSM algorithm is efficient, and is excellent at handling non-Gaussian noise.

In this paper, a novel Takagi-Sugeno (T-S) fuzzy model particle filtering algorithm (TSF-PF) based on fuzzy C-regression clustering is proposed for uncertainty modeling of the target dynamic model with non-Gaussian noise. In the proposed algorithm, a generic semantic framework of the T-S fuzzy model is constructed to incorporate spatial feature information of a target into the particle filter, in which the spatial feature information is characterized by several semantic fuzzy sets. Meanwhile, a fuzzy C-regression clustering method based on correntropy is proposed to adaptively identify the premise parameters of the T-S fuzzy model, which is used to adjust the weight of models, and a Kalman filter is used to identify the consequent parameters. And then an efficient importance density function is constructed by using the proposed T-S fuzzy model, which can efficiently improve the robust and diversity of the sampling particles. Furthermore, in order to improve the real-time performance of the proposed algorithm, two improved T-S fuzzy model particle filtering algorithms are presented. The simulation results show that the tracking performance of the proposed algorithms are better than that of the traditional interacting multiple model (IMM), interacting multiple model unscented Kalman filter (IMMUKF), interacting multiple model particle filter (IMMPF) and interacting multiple model Rao-Blackwellized particle filter (IMMRBPF). Particularly, the proposed algorithms can accurately track the maneuvering target when the moving direction abruptly changes or the prior information of the target dynamic model is inaccuracy.

In this paper, a novel intuitionistic fuzzy clustering algorithm based on feature selection (IFC-FS) for multiple object tracking is proposed. In the proposed algorithm, the neighborhood rough set is used to achieve the adaptive selection of the multiple object features of visual objects, which are applied to calculate the distance similarity measure between the objects and the observations. At the same time, in order to incorporate the local information of objects into the intuitionistic fuzzy clustering, the local information distances between objects and observations are estimated by using the optimal subpattern assignment metric based on the reference topology set, and a new intuitionistic fuzzy clustering based on maximum entropy principle is proposed by using the new similarity distance measure. Finally, the association probabilities among the objects and the observations are reconstructed by utilizing the intuitionistic fuzzy membership degrees. The experimental results show that the proposed algorithm can effectively improve the estimated accuracy and robustness of the association probabilities between the objects and the observations, and have the ability to track accurately multiple objects in the complex background and long-time occlusion environment.

In this paper, we propose a novel recursive Takagi-Sugeno (T-S) fuzzy semantic modeling approach for discrete state-space system. According to the information learning theoretic (ILT), the correntropy can capture the higher moments of the error probability distribution to deal with non-Gaussian noise. Considering the advantages of fuzzy theory and correntropy, fuzzy correntropy is constructed and a novel kernel fuzzy C-regression model clustering based on fuzzy correntropy is proposed to solve the premise parameter identification problem of the T-S fuzzy model. To the identification of the consequent part parameters of the T-S fuzzy model, a modified extended forgetting factor recursive least squares (MEFRLS) estimator is presented. Moreover, to evaluate the performance of the proposed fuzzy model, the proposed T-S fuzzy model is applied to solve the problem of maneuvering target tracking by incorporating the target feature semantic information. Finally, the experiment results show the proposed algorithm can effectively track a maneuvering target, and its performance is better than the exist algorithms, such as interacting multiple model Kalman filter (IMMKF), interacting multiple model unscented Kalman filter (IMMUKF), the interacting multiple model particle filter (IMMPF) and interacting multiple model Rao-Blackwellized particle filter the (IMMRBPF).

The probability hypothesis density filter under glint noise is efficient for multi-object tracking. However, this filter is inapplicable to tracking the multiple maneuvering objects in case of low detection probability. In order to track the multiple maneuvering objects under glint noise more efficiently, we propose a novel multi-object Bayesian (MOB) filter for jump Markov system under glint noise by applying jump Markov system models and variational Bayesian approach to the MOB filter and develop an implementation of this filter. The developed implementation uses the Student's t-distribution to depict the glint noise, uses the Gaussian-Gamma distribution to represent the predicted state distribution and updated state distribution of each target, and uses the variational Bayesian method to acquire the approximate state distribution of each target. A comparison of the proposed MOB filter with the existing filters demonstrates that the proposed MOB filter is better at tracking the maneuvering objects under glint noise than the existing filters.

Considering the development of object detection based on deep learning framework in recent years, it has brought a new scope for multi-source fusion in the field of autonomous driving. In this paper, we propose a new architecture with a feature pyramid attention module to fuse the projected radar data and camera data. In the proposed algorithm, the detection model of YOLOv3 is employed by us and the feature pyramid module is extended with the input interface of the radar projection image and attention module. Additionally, in order to reduce the interference information from different scales of radar projected block, a new generation mechanism of radar projection images is introduced. Finally, the radar projection image is fused in feature pyramid layers with an attention module. The result shows that the proposed fusion algorithm outperforms better than image-only network for the nuScenes dataset. The code for this research will be made available to the public at: https://github.com/yuanliangxie/nuscenes_data_process.

This paper presents an effective constrained multiple model particle filtering (CMMPF) for bearings-only maneuvering target tracking. In the proposed algorithm, the process of target tracking is factorized into two sub-problems: 1) motion model estimation and model-conditioned state filtering according to the Rao–Blackwellised theorem and 2) the target dynamic system is modeled by multiple switching dynamic models in a jump Markov system framework. To estimate the model set, a modified sequential importance resampling method is used to draw the model particles, which can be restricted into the feasible area coincide with the constrained bound. To the model-conditioned state nonlinear filtering, a truncated prior probability density function is constructed by utilizing the latest observations and auxiliary variables (target spatio–temporal features), which can guarantee the diversity and accuracy of the sampled particles. The tracking performance is compared and analyzed with other conventional filters in two scenarios: 1) uniform and time-invariant sampling scenario and 2) non-uniform and sparse sampling scenario. A conservative Cramer–Rao lower bound is also introduced and compared with the root mean square error performance of the suboptimal filters. Simulation results confirm the superiority of CMMPF algorithm over the other existing ones in comparison with respect to accuracy, efficiency, and robustness for the bearings-only target tracking system, especially for the aperiodic and sparse sampling environment.

This paper proposes a marginal multi-object Bayesian filter with multiple hypotheses to track multiple objects in the presence of object appearing and object disappearing, missed detection and clutter. This filter delivers the probability of existence and probability density function of each object. A mathematical model for searching K-best hypotheses is set up by the maximization of the generalized joint likelihood ratios of hypotheses, which results in a 2-dimensional assignment problem. The K-best hypotheses can be acquired by using the Murty algorithm to solve the 2-dimensional assignment problem. According to the K-best hypotheses, the existence probabilities and probability density functions of objects are formed. Furthermore, an implementation of this filter for a linear Gaussian system is developed and is extended to nonlinear observations. Experimental result demonstrates that the proposed filter outperforms other available filters at various numbers of clutter and different detecting probabilities.

The probability hypothesis density (PHD) filter and marginal distribution Bayes (MDB) filter are two efficient Bayes approaches for multi-target tracking. However, these two filters fail to provide the state estimation of a target during its initial times due to the poor capability of the two filters on the target detection. To enhance the capability of the MDB filter on the target detection, we present a method for the target detection based on the rule-based track initiation technique, and develop a multi-target Bayes filter with the target detection by applying this target detection method to the MDB filter. Simulation results indicate that this filter has a stronger detecting and tracking capability of the target than the existing PHD and MDB filters.

Aiming at the modeling uncertainty and nonGaussian characteristics of dynamic systems in maneuvering target tracking, a Takagi-Sugeno (T-S) modeling particle filter method via intuitionistic fuzzy decision (IFD-TSPF) is proposed. As for the Structure recognition of T-S model, the membership degree of multiple features are calculated by the ridge reduction distribution function, and then the intuitionistic fuzzy point operator is used to calculate the decision scores corresponding to the features. According to the principle of evidence decision, the appropriate premise variables are selected. Finally, the importance density function is constructed by the estimation results of the proposed T-S model, which can effectively reduce the particle degradation. Through explanation of simulation results, which show that the performance of our algorithm is better than interacting multi-model particle filter(IMMPF) and interacting multi-model Rao-Blackwellized particle filter(IMMRBPF). When the direction of the target is suddenly changed, it can effectively track the target.

In this paper, we propose a novel fuzzy expectation maximization (FEM) based Takagi-Sugeno (T-S) fuzzy particle filtering (FEMTS-PF) algorithm for a passive sensor system. In order to incorporate target spatial-temporal information into particle filtering, we introduce a T-S fuzzy modeling algorithm, in which an improved FEM approach is proposed to adaptively identify the premise parameters, and the model probability is adjusted by the premise membership functions. In the proposed FEM, the fuzzy parameter is derived by the fuzzy C-regressive model clustering method based on entropy and spatial-temporal characteristics, which can avoid the subjective influence caused by the artificial setting of the initial value when compared to the traditional FEM. Furthermore, using the proposed T-S fuzzy model, the algorithm samples particles, which can effectively reduce the particle degradation phenomenon and the parallel filtering, can realize the real-time performance of the algorithm. Finally, the results of the proposed algorithm are evaluated and compared to several existing filtering algorithms through a series of Monte Carlo simulations. The simulation results demonstrate that the proposed algorithm is more precise, robust and that it even has a faster convergence rate than the interacting multiple model unscented Kalman filter (IMMUKF), interacting multiple model extended Kalman filter (IMMEKF) and interacting multiple model Rao-Blackwellized particle filter (IMMRBPF).

For the problem of inaccurate or difficult to obtain statistical characteristics of non-Gaussian noise, an interacting T-S fuzzy modeling algorithm is proposed to incorporate spatial-temporal information into particle filtering. In the proposed method, feature information is characterized by multiple semantic fuzzy sets, and the model transition probabilities are estimated by using the fuzzy set transition probabilities, which can be derived by the closeness degrees between the fuzzy sets. Furthermore, the correntropy can capture the statistical information to solve the non-Gaussian noise, thus a kernel fuzzy C-regression means (FCRM) based on correntropy and spatial-temporal information is proposed to adaptively identify the premise parameters of T-S fuzzy model, and a modified strong tracking method is used to estimate the consequence parameters. By using the proposed interacting T-S fuzzy model, an efficient importance density function is constructed for the particle filtering algorithm. Finally, the simulation results show that the tracking performance of the proposed algorithm is effective in processing non-Gaussian noise.

Tracking multiple maneuvering (MM) targets is a well-known and challenging problem because of clutter and several uncertainties existing in target motion mode, target detection, and data association. An efficient solution to this problem is the Gaussian mixture probability hypothesis density (GM-PHD) filter for jump Markov system (JMS) models. However, this solution is inapplicable to circumstances where detection probability is low because the GM-PHD filter for JMS models requires a high detection probability. To address this problem, we propose a sequential multiple target (MT) Bayes filter for JMS models. To track MM targets that are switching among a set of linear Gaussian models, an implementation process of this filter for linear Gaussian jump Markov MT models is also developed. The conclusion that the novel filter is more efficient for tracking MM targets than the existing filter for JMS models in circumstances of low detection probability is validated by simulation results.