Jason L. Williams

Jason L. Williams

Recent developments in random finite sets (RFSs) have yielded a variety of tracking methods that avoid data association. This paper derives a form of the full Bayes RFS filter and observes that data association is implicitly present, in a data structure similar to MHT. Subsequently, algorithms are obtained by approximating the distribution of associations. Two algorithms result: one nearly identical to JIPDA, and another related to the MeMBer filter. Both improve performance in challenging environments.

Jason L. Williams

Random finite sets (RFSs) has been a fruitful area of research in recent years, yielding new approximate filters such as the probability hypothesis density (PHD), cardinalised PHD (CPHD), and multiple target multi-Bernoulli (MeMBer). These new methods have largely been based on approximations that side-step the need for measurement-to-track association. Comparably, RFS methods that incorporate data association, such as Morelande and Challa's (M-C) method, have received little attention. This paper provides a RFS algorithm that incorporates data association similarly to the M-C method, but retains computational tractability via a recently developed approximation of marginal association weights. We describe an efficient method for resolving the track coalescence phenomenon which is problematic for joint probabilistic data association (JPDA) and related methods (including M-C). The method utilises a network flow optimisation, and thus is tractable for large numbers of targets. Finally, our derivation also shows that it is natural for the multi-target density to incorporate both a Poisson point process (PPP) component (representing targets that have never been detected) and a multi-Bernoulli component (representing targets under track). We describe a method of recycling, in which tracks with a low probability existence are transferred from the multi-Bernoulli component to the PPP component, effectively yielding a hybrid of M-C and PHD.

Ángel F. García-Fernández, Jason L. Williams, Lennart Svensson et al.

This paper proposes a Poisson multi-Bernoulli mixture (PMBM) filter for coexisting point and extended targets, i.e., for scenarios where there may be simultaneous point and extended targets. The PMBM filter provides a recursion to compute the multi-target filtering posterior based on probabilistic information on data associations, and single-target predictions and updates. In this paper, we first derive the PMBM filter update for a generalised measurement model, which can include measurements originated from point and extended targets. Second, we propose a single-target space that accommodates both point and extended targets and derive the filtering recursion that propagates Gaussian densities for point targets and gamma Gaussian inverse Wishart densities for extended targets. As a computationally efficient approximation of the PMBM filter, we also develop a Poisson multi-Bernoulli (PMB) filter for coexisting point and extended targets. The resulting filters are analysed via numerical simulations.

Ángel F. García-Fernández, Lennart Svensson, Jason L. Williams et al.

This paper presents two trajectory Poisson multi-Bernoulli (TPMB) filters for multi-target tracking: one to estimate the set of alive trajectories at each time step and another to estimate the set of all trajectories, which includes alive and dead trajectories, at each time step. The filters are based on propagating a Poisson multi-Bernoulli (PMB) density on the corresponding set of trajectories through the filtering recursion. After the update step, the posterior is a PMB mixture (PMBM) so, in order to obtain a PMB density, a Kullback-Leibler divergence minimisation on an augmented space is performed. The developed filters are computationally lighter alternatives to the trajectory PMBM filters, which provide the closed-form recursion for sets of trajectories with Poisson birth model, and are shown to outperform previous multi-target tracking algorithms.

Ángel F. García-Fernández, Yuxuan Xia, Karl Granström et al.

This paper presents the Gaussian implementation of the multi-Bernoulli mixture (MBM) filter. The MBM filter provides the filtering (multi-target) density for the standard dynamic and radar measurement models when the birth model is multi-Bernoulli or multi-Bernoulli mixture. Under linear/Gaussian models, the single target densities of the MBM mixture admit Gaussian closed-form expressions. Murty's algorithm is used to select the global hypotheses with highest weights. The MBM filter is compared with other algorithms in the literature via numerical simulations.

Ángel F. García-Fernández, Jason L. Williams, Karl Granström et al.

We provide a derivation of the Poisson multi-Bernoulli mixture (PMBM) filter for multi-target tracking with the standard point target measurements without using probability generating functionals or functional derivatives. We also establish the connection with the δ-generalised labelled multi-Bernoulli (δ-GLMB) filter, showing that a δ-GLMB density represents a multi-Bernoulli mixture with labelled targets so it can be seen as a special case of PMBM. In addition, we propose an implementation for linear/Gaussian dynamic and measurement models and how to efficiently obtain typical estimators in the literature from the PMBM. The PMBM filter is shown to outperform other filters in the literature in a challenging scenario.

Jason L. Williams, Roslyn A. Lau

Data association, the reasoning over correspondence between targets and measurements, is a problem of fundamental importance in target tracking. Recently, belief propagation (BP) has emerged as a promising method for estimating the marginal probabilities of measurement to target association, providing fast, accurate estimates. The excellent performance of BP in the particular formulation used may be attributed to the convexity of the underlying free energy which it implicitly optimises. This paper studies multiple scan data association problems, i.e., problems that reason over correspondence between targets and several sets of measurements, which may correspond to different sensors or different time steps. We find that the multiple scan extension of the single scan BP formulation is non-convex and demonstrate the undesirable behaviour that can result. A convex free energy is constructed using the recently proposed fractional free energy (FFE). A convergent, BP-like algorithm is provided for the single scan FFE, and employed in optimising the multiple scan free energy using primal-dual coordinate ascent. Finally, based on a variational interpretation of joint probabilistic data association (JPDA), we develop a sequential variant of the algorithm that is similar to JPDA, but retains consistency constraints from prior scans. The performance of the proposed methods is demonstrated on a bearings only target localisation problem.

Jason L. Williams

The joint probabilistic data association (JPDA) filter is a popular tracking methodology for problems involving well-spaced targets, but it is rarely applied in problems with closely-spaced targets due to its complexity in these cases, and due to the well-known phenomenon of coalescence. This paper addresses these difficulties using random finite sets (RFSs) and variational inference, deriving a highly tractable, approximate method for obtaining the multi-Bernoulli distribution that minimises the set Kullback-Leibler (KL) divergence from the true posterior, working within the RFS framework to incorporate uncertainty in target existence. The derivation is interpreted as an application of expectation-maximisation (EM), where the missing data is the correspondence of Bernoulli components (i.e., tracks) under each data association hypothesis. The missing data is shown to play an identical role to the selection of an ordered distribution in the same ordered family in the set JPDA algorithm. Subsequently, a special case of the proposed method is utilised to provide an efficient approximation of the minimum mean optimal sub-pattern assignment estimator. The performance of the proposed methods is demonstrated in challenging scenarios in which up to twenty targets come into close proximity.

Jason L. Williams, Roslyn A. Lau

Data association, the problem of reasoning over correspondence between targets and measurements, is a fundamental problem in tracking. This paper presents a graphical model formulation of data association and applies an approximate inference method, belief propagation (BP), to obtain estimates of marginal association probabilities. We prove that BP is guaranteed to converge, and bound the number of iterations necessary. Experiments reveal a favourable comparison to prior methods in terms of accuracy and computational complexity.

Jason L. Williams

Recent developments in random finite sets (RFSs) have yielded a variety of tracking methods that avoid data association. This paper derives a form of the full Bayes RFS filter and observes that data association is implicitly present, in a data structure similar to MHT. Subsequently, algorithms are obtained by approximating the distribution of associations. Two algorithms result: one nearly identical to JIPDA, and another related to the MeMBer filter. Both improve performance in challenging environments.

Jason L. Williams

The probability hypothesis density (PHD) and multi-target multi-Bernoulli (MeMBer) filters are two leading algorithms that have emerged from random finite sets (RFS). In this paper we study a method which combines these two approaches. Our work is motivated by a sister paper, which proves that the full Bayes RFS filter naturally incorporates a Poisson component representing targets that have never been detected, and a linear combination of multi-Bernoulli components representing targets under track. Here we demonstrate the benefit (in speed of track initiation) that maintenance of a Poisson component of undetected targets provides. Subsequently, we propose a method of recycling, which projects Bernoulli components with a low probability of existence onto the Poisson component (as opposed to deleting them). We show that this allows us to achieve similar tracking performance using a fraction of the number of Bernoulli components (i.e., tracks).