ABSTRACT. A key challenge for multi-target trackers is being able to track both agile and simply moving targets effectively. This is only furthered by the standard difficulties of automated track initiation and deletion. This paper proposes a solution to this, where unknown associations and track management are handled by a Dirichlet process prior, and Gaussian processes model the dynamics of the targets. The promising performance of the proposed tracker is demonstrated on both synthetic and real radar data against a selection of other methods.

ABSTRACT. The aeroelastic behaviour of aircraft is parameter variant. Changing flight conditions, such as e.g. flight velocity and altitude may change the vibration damping. When the vibration damping becomes zero or negative, self-excitation of the vibration occurs, called flutter. Modal parameter identification can be applied to extract eigenfrequencies and damping ratios based on e.g. acceleration data. In order to avoid flutter, modal parameters can be identified in flight testing of a new aircraft type close to real-time using optimized algorithms. Real-time identification of modal parameters has significant uncertainties, especially with respect to damping ratios. Those uncertainties cannot be calculated, but qualitatively estimated. In this study, a Kalman filter tracking is applied to reduce the uncertainties of modal parameter monitoring of aircraft. Since the process noise of such a system is impossible to foresee and is expected to change throughout a flight, the process noise is adapted with respect to the innovation and changing flight conditions. This contextaware adaptive Kalman filter is tested on data from a simulated aeroelastic model as well as on real flight test data of a smallscale fixed-wing UAV. The results show significant reduction of the identification uncertainties for both simulated and real data.

ABSTRACT. The paper first derives expressions for the MSE and bias of a mismatched linear filter applied to an arbitrary trajectory and describes how such expressions can be used for process noise selection. The performance of such algorithms when applied to the selection of process noise terms for a 1D sinusoidal dynamic model and for a 2D weaving dynamic model is then considered. Previously, the only algorithm specifically designed for process noise selection of 1D sinusoidal models is a 1995 paper by Sudano. Optimal gains are compared with suboptimal techniques in 1D and 2D, including with Sudano's algorithm. It is shown that the algorithm optimizing over the explicit MSE outperforms the other techniques both in minimizing the MSE and also in guaranteeing that the peak error is never worse than "connecting the dots" between measurements.

ABSTRACT. In Multiple Extended Object Tracking, the PMBM (Poisson Multi-Bernoulli Mixture) tracker is considered state-of-the-art. Originally, it was presented with the GGIW (Gamma Gaussian Inverse Wishart) target model, which is a random matrix model. When tracking larger objects using LiDAR, measurements are generated by the contour rather than the whole target surface, and it is beneficial to model this with the target model. A target model which has this capability is the Gaussian Process (GP) extent model. This paper presents a PMBM tracker using this target model. We also discuss considerations related to the use of the GP model in the PMBM framework. Secondly, we present improvements in the target model which increases the robustness of the model by dealing with the inherent non-linearities using the Gauss-Newton method. We also present a comparison with the GGIW-PMBM tracker on simulated and real LiDAR data gathered from maritime vessels.

ABSTRACT. This paper presents an efficient implementation of the trajectory Poisson multi-Bernoulli (PMB) filter for multiple extended object tracking (EOT), which directly estimates a set of object trajectories. The trajectory PMB filter propagates a PMB density on the posterior of sets of trajectories through the filtering recursions over time, where the multi-Bernoulli (MB) mixture in the PMB mixture (PMBM) posterior after each update step is approximated as a single MB. The efficient MB approximation is achieved by first running a blocked Gibbs sampler on the joint posterior of the set of trajectories and the measurement association variables. The single-object measurement model is assumed to be a Poisson point process which enables us to parallelize the sampling across all objects and association variables, respectively. Then, samples of object states are utilized to form the approximate MB density via Kullback-Leibler divergence minimization. Simulation results on EOT with known and constant elliptical shapes show that the TPMB implementation using blocked Gibbs sampling outperforms the state-of-the-art TPMB implementation using loopy belief propagation with significantly reduced runtime.

ABSTRACT. In this work, the multiple GP-based filter (MGP) is proposed for tracking an extended target with a random shape. This model is able to represent one or a group of targets, where the shape of each sub-target is described by the GP model. Utilizing the MGP method for extended target tracking, the mainly challenge are as follows: (1) the data association problem of the inter-cluster; (2) the challenge of approximating nonconjugate distributions. In order to solve the association problem of the measurement with the sub-targets, a latent variable is defined to correlate them. Due to the lack of conjugacy, the resulting posterior is not possible to find a closed-form analytical expression. This paper uses variational inference to solve the problem and approximate the true posterior. The effectiveness of the proposed algorithm is evaluated through simulation.

ABSTRACT. A generalisation of the Kalman filter for random fields is proposed. The result exploits a connection between Gibbs point processes and Gaussian random fields. This result provides a solution to the multi-object filtering problem in terms of the first two moments of a Gaussian random field. Applications are discussed for different multi-target tracking models.

ABSTRACT. Gibbs sampling is one of the most popular Markov chain Monte Carlo algorithms because of its simplicity, scalability, and wide applicability within many fields of statistics, science, and engineering. In the labeled random finite sets literature, Gibbs sampling procedures have recently been applied to efficiently truncate the single-sensor and multi-sensor delta-generalized labeled multi-Bernoulli posterior density as well as the multi-sensor adaptive labeled multi-Bernoulli birth distribution. However, only a limited discussion has been provided regarding key Gibbs sampler architecture details including the Markov chain Monte Carlo sample generation technique and early termination criteria. This paper begins with a brief background on Markov chain Monte Carlo methods and a review of the Gibbs sampler implementations proposed for labeled random finite sets filters. Next, we propose a short chain, multi-simulation sample generation technique that is well suited for these applications and enables a parallel processing implementation. Additionally, we present two heuristic early termination criteria that achieve similar sampling performance with substantially fewer Markov chain observations. Finally, the benefits of the proposed Gibbs samplers are demonstrated via two Monte Carlo simulations.

ABSTRACT. Measurement-adaptive track initiation remains a critical design requirement of many practical multi-target tracking systems. For labeled random finite sets multi-object filters, prior work has been established to construct a labeled multi-object birth density using measurements from multiple sensors. A truncation procedure has also been provided that leverages a stochastic Gibbs sampler to truncate the birth density for scalability. In this work, we introduce a deterministic herded Gibbs sampling truncation solution for efficient multi-sensor adaptive track initialization. Removing the stochastic behavior of the track initialization procedure without impacting average tracking performance enables a more robust tracking solution more suitable for safety-critical applications. Simulation results for both linear and non-linear sensing scenarios are provided to verify performance.

ABSTRACT. The shifted Dirichlet distribution is utilized to extend the definition of subjective opinions to include upper and lower bounds on belief and base rates. The characteristics of the bounded subjective opinions are examined and contrasted with unbounded subjective opinions.

ABSTRACT. Operation monitoring for automation systems requires self-assessment of all data processing modules. In this work, we extend our new self-assessment method for linear Kalman filters based on subjective logic to nonlinear Kalman filtering. Furthermore, we propose novel approaches within this subjective logic-based framework to assess the overall filter performance in multi-sensor systems online, i.e., in real-time without ground truth data. The results of the proposed self-assessment method for nonlinear Kalman filtering are demonstrated through simulation studies, showing advantages compared to classical consistency measures, like the normalized innovation squared. In addition, the results of the proposed online overall filtering assessment for multi-sensor systems can even compete with consistency measures based on ground truth data, which cannot be applied in online applications.

ABSTRACT. The already highly complex Urban Transit Routing Problem (UTRP) that serves to find efficient travelling routes for Public Transport (PT) systems is extended into the Heterogeneous Electric - Urban Transit Routing Problem (HE-UTRP). This extension focuses on step-by-step transformation of public bus transportation systems to electric mobility. The heterogeneity characteristics refers to the fleet and charging infrastructure. This article presents a framework that allows the generation, analysis and optimisation of PT Route Network (RN) for the HE-UTRP. In addition to the analysis of different charging technologies and Charging Locations, the approach enables a transformation process towards electrification of PT systems by presenting substitution scenarios as well as the resulting cost structure. The framework, based on a Sequence-based Selection Hyper-heuristic - with Great Deluge, is tested against varying objective functions and UTRP, HE-UTRP and Electric Transit Route Network Design Problem instances.

ABSTRACT. We aim to approximate the distribution of the first time a particle moving according to a Gaussian process with increasing trend, e.g., a state space model such as a constant-velocity (CV) or constant-acceleration (CA) model, arrives at a fixed location. Motivated by an engineering particle transport problem for which we can assume that once a particle has arrived at this location it cannot move back, we derive a simple, analytic approximation for the first-passage time probabilities and calculate its inverse cumulative distribution function analytically and the moments numerically. Furthermore, we propose a simple Gaussian approximation based on a linearization approach. The strengths and limitations of our methods are discussed, and by comparison with Monte Carlo simulations, we show that in particular, the first one satisfies the requirements of engineering problems in terms of accuracy and computation time.

ABSTRACT. We consider the problem of real-time estimation of sea state and wave-induced motions on a moving vessel using onboard inertial sensors without knowing vessel's dynamic parameters (i.e., draught and breadth). This is crucial for vessel operational planning and performance, preventing structure failure, emissions reduction and fuel economy. This work proposes a new estimation approach by reformulating the conventional problem of sea state and vessel motion estimation (unknown input into a known dynamic system) as an input-state-parameter estimation problem of mass-spring-damper systems. We exploit the strong correlations between a vessel's vertical displacement and its rotation to develop a new estimation algorithm---Parameter-Sharing Extended-Augmented Kalman Filter (PS-EAKF)---for the problem to estimate the unidentified vessel parameters together with vessel motion (heave and pitch) and sea state. Experimental data from a scale-model vessel in regular head seas demonstrate the effectiveness and robustness of the proposed approach.

ABSTRACT. The increasing advances in Unmanned Systems are transforming the type of threats traditional defence systems are designed to tackle. One significant advancement is in artificial intelligence capability which allows a group of agents to perform complex collective behaviors. As consequence, providing defence systems with threat intelligence capability is becoming a necessity, where identifying collective behaviours and coordinated targets can significantly increase the effectiveness of the system. In this work, we propose a probabilistic approach to identify certain types of coordinated targets. A scoring function that utilizes accumulated state densities (ASDs) over a sliding time window is proposed to compute the likelihood of a pair of targets being coordinated. A simulation scenario including different types of coordination is used to test the performance of this proposed method.

ABSTRACT. This paper announces version 1.0 of Stone Soup: the open-source tracking and state estimation framework. We highlight key elements of the framework and outline example applications and community activities.

Stone Soup is engineered with modularity and encapsulation at its heart. This means that its many components can be put together in any number of ways to build, compare, and assure almost any type of multi-target tracking and fusion algorithm. Since its inception in 2017, it has aimed to provide the target tracking and state estimation community with an open, easy-to-deploy framework to develop and assess the performance of different types of trackers. Now, through repeated application in many use cases, implementation of a wide variety of algorithms, multiple beta releases, and contributions from the community, the framework has reached a stable point.

In announcing this release, we hope to encourage additional adoption and further contributions to the toolkit. We also acknowledge and express appreciation for the many contributions of time and expertise donated by the tracking and fusion community.

ABSTRACT. This paper introduces the concept of the Gaussian integral filter (GIF), the limit of the Gaussian sum filter (GSF) for when the number of mixands tends to infinity. The GIF is obtained via a combination of GSF, quadrature, and interpolation. While it is a very general concept, in this paper the GIF is used to represent multiviariate Laplace (ML) distributions defining the process noise when tracking a maneuvering target. The filter is first applied to a linear three-dimensional toy problem, and then to a maneuvering target tracking problem in Earth orbit. For the more complex maneuvering target tracking problem, the filter requires only 1.4 times the computational resources of an unscented Kalman filter (UKF), while having errors up to 11 times smaller. For the same problem, the UKF slowly diverges.

ABSTRACT. With the growing availability of high-resolution sensors, processing more than one detection per target becomes increasingly critical when tracking multiple extended objects. However, contemporary sensors often generate spurious detections that need to be considered. Naively employing standard multitarget trackers may result in poor tracking performance for multitarget-multidetection tracking in cluttered environments, and the relevant extensions are nontrivial.This paper introduces a version of the kernel symmetric measurement equation (SME) filter that considers both multidetections and clutter. For a simulated scenario, our novel filter achieved a higher accuracy than the global nearest neighbor (GNN) and a fast variant of the joint probabilistic data association filter (JPDAF).

ABSTRACT. Recursive estimation of the orientation and spatial extent of an object from sparse measurements is a challenging yet crucial problem. To this end, we propose a problem-tailored particle filter that employs a special importance density  for efficient sampling from high likelihood regions. This importance density is obtained with the help of a suitable analytic method for determining the length and width of the object. Due to the geometric meaning of the state vector, a consolidated estimate from the individual particles is obtained based on the Gaussian Wasserstein distance. The resulting filter can be employed with different analytic methods, and is found to improve estimation accuracy beyond state-of-the-art algorithms in a challenging scenario. Especially for low measurement rates, the proposed filter yields improved results compared to the reference methods.

ABSTRACT. Successful tracking of an extended object requires accurate estimation of the target's shape. In order to evaluate the estimation performance of such tracking algorithms, the target shape must be incorporated by a suitable metric or score. In this context, a common task is to determine a consolidated scalar score for the estimation accuracy across a set of different target types. We highlight problems exhibited by existing scores in this case, particularly if the data set consists of objects of diverse sizes. To this end, we focus on elliptical targets. Furthermore, we present a scale-invariant adaption of the commonly used Gaussian Wasserstein Distance, which does not suffer from the highlighted problem.

ABSTRACT. Multi-object filtering is a generalisation of stochastic filtering to deal with an unknown and time-varying number of targets, largely based on modelling with point processes. Some early works on this topic from the Soviet Union from 1960s-1980s appeared prior to well known results in the contemporary literature This article reviews some of these historical contributions.

ABSTRACT. The main challenge in random finite set-based multi-sensor multi-object tracking is the NP-hard association of the sensor measurements with the tracks. Using the Bayes parallel combination rule, we have recently proposed the Product Multi-sensor Generalized Labeled Multi-Bernoulli (PM-GLMB) filter, decomposing the multi-sensor update into simpler single-sensor problems with subsequent Bayes optimal fusion. This paper extends the PM-GLMB filter to prior densities with Gaussian mixture spatial distributions, which is an inevitable prerequisite for the Product Multi-sensor Labeled Multi-Bernoulli (PM-LMB) filter we propose afterward. Finally, we evaluate the performance of both filters in a simulation study. In this context, we address the known performance problems of the PM-GLMB filter in crowded situations and show how the PM-LMB filter overcomes these.

ABSTRACT. The multi-sensor Labeled Multi-Bernoulli filter has the challenge of relying on the NP-hard multi-sensor update of the Generalized Labeled Multi-Bernoulli filter. This paper proposes the Fast Product Multi-Sensor Labeled Multi-Bernoullifilter, which is a filter for multi-sensor systems that solves this task by performing computationally simpler single-sensor LabeledMulti-Bernoulli filter updates based on a common prediction for each sensor. These single-sensor updates are then fused using a novel and efficient fusion strategy. Furthermore, the proposed filter is based on the Bayes parallel combination rule and can be seen as an efficient approximation of the multi-sensor LabeledMulti-Bernoulli filter. It enables full parallelization of the update step and benefits from sensor order independence compared toIterated Corrector implementations. As a result, the robustness is increased, which is important for safety reasons, e.g., in autonomous driving. Our approach is evaluated on simulations, and the results are compared to an Iterated Corrector implementation of the Labeled Multi-Bernoulli filter.

ABSTRACT. Epistemic decisions about which sources to trust and query are critical for a decision-maker, when the end-goal (or final) decisions are to be made using limited resources. Toward this, we previously proposed preliminary extensions to classical measures of Value of Information (VoI) for imprecise belief states represented by belief functions relying on a general observation model. These methods primarily aim to assist a decision-maker toward making rational decisions, by generating necessary metadata about the information that is being considered for the decision-making task (e.g. probability of source reliability, degree of self-confidence expressed by the source). The use of such measures is even more critical in a multi-intelligence context, where information sources are highly heterogeneous, prone to errors, partially informative, deceptive, or even malicious. In this paper, we explore the behavior and performance of our previously proposed belief theoretic VoI measures, the Evidential Expecetd Value of Sampled Information (EEVSI), and propose a procedure to sequentially select sources to query. We also consider the case where the information sources are providing contradictory evidence. We leverage a maritime surveillance scenario, where the decision-maker has to make a rational ordered selection of information sources among a set consists of both physical sensors and human sources, to illustrate the behavior of the proposed method. We conclude by providing some insights on future research directions to further expand on our proposed new measures

ABSTRACT. Human Activity Recognition (HAR) based on wearable device has become as a hot topic of research due to its wide range of applications in health-care, fitness and smart homes. However, the classification of some activities with similar sensor readings, such as standing and sitting, is usually more challenging for the design of efficient activity recognition algorithms. Considering the inconsistent performance of different classifiers, which can provide information complementary for individual classifier, we propose a novel multi-classifier fusion method based on belief functions (BFs) theory for HAR. Specifically, at first, four classifiers are trained using time-domain and frequency-domain features to obtain basic belief assignments (BBA) of activity, respectively. Then, three assessment criteria are utilized to evaluate the reliability of the classifiers and a scoring matrix is constructed. Next, the algorithm of Belief Function based the Technique for Order Preference by Similarity to Ideal Solution (BF-TOPSIS) is employed to calculate the weighting coefficients for each classifier. Finally, the discounting and Dempster’s rules are adopted to combine the multiple classifiers and further decision making. Several experiments were conducted to illustrate the performance of the proposed method using the UCI smartphone dataset, and the results show that the proposed method is more accurate than the state-of-art methods.

ABSTRACT. Wearable smart devices are capable of capturing a variety of information from their users using a multitude of noninvasive sensing modalities. Using features from the raw measurements of wearable devices, sensor fusion enables us to obtain a holistic picture of the users' context and monitor their activity state with increased accuracy. Human activity recognition using noninvasive sensors allows us to capture the natural behavior of users in their day-to-day lives. This in-the-wild activity recognition, however, poses several key challenges that must be addressed to create effective classification models. The main challenges are class imbalance, uncertainty in classifier decisions, and large feature spaces. To address them, this study further explores a probabilistic sensor fusion method called Naive Adaptive Probabilistic Sensor (NAPS) Fusion. In doing so, we establish the viability of NAPS Fusion for natural human activity recognition using noninvasive sensing modalities. NAPS Fusion handles dimensionality reduction by creating reduced feature sets and mitigates the class imbalance issue through the use of Synthetic Minority Oversampling Technique (SMOTE). Moreover, NAPS Fusion addresses uncertainty in the decisions of classifiers using a Dempster-Shafer theoretic late fusion framework. Our empirical evaluation demonstrates that NAPS Fusion has broad applications beyond its original design for cognitive state detection. It outperforms similar decision level sensor fusion methods (late fusion using averaging, LFA, and late fusion using learned weights, LFL) in the detection of exercise and sedentary activities such as walking, running, lying down, and sitting. We observe improvements of up to 56% in F1 score and up to 59% in precision with NAPS Fusion over the compared methods.

ABSTRACT. In a context of autonomous driving, perception of the surrounding is a crucial task. It characterizes the vehicle’s ability to simultaneously model its surroundings accurately and maintain its position in the environment. In this article, a new framework of mobile robot perception and risk assessment is proposed. Our approach aims to leverage the simultaneous combination of the standard occupancy grid map with a new map that we have called ”knowledge map”. This proposal was motivated by the fact that risk arises not only from obstacles but also from the lack of knowledge. Using this framework, we are able to assess the risk, mainly of collision, over a given path P and therefore compute an optimal navigation control of the robot. Thanks to the proposed Bayesian framework the paper also shows how we can combine both local measurements and existing map (eg. OpenStreetMap) and also take account of the robot’s localization errors.

ABSTRACT. Military systems require accurate technical solutions to support decision making. In naval warfare, an important problem to solve relies on a capability to detect higher level category artifacts, such as warship formations. This work aims to fill a gap observed in literature about this subject. In this paper, we proposed, implemented and tested a model to detect warship formations, capable of classifying the formation according to its type. We also investigate some published works related to this subject and point out the differences and gaps perceived. Using a Kohonen Network as classifier based on position and velocity of ships, this work describes the computational model used and the results obtained according to different number of ships and samples. Using synthetic data combined with real data, the results show, with different types of metrics, that the adopted solution has reliable and promising results and it is adequate to deal with a real-world problem.

ABSTRACT. Within the EUROCONTROL Air Traffic Management (ATM) architecture, the Surveillance Analysis Support System for ATC Centres (SASS-C) is dedicated to the validation of Air Navigation Service Providers. One of its applications is the Opportunity Traffic Reconstruction (OTR), which seeks to reconstruct all trajectories by combining noisy measurements from different sensors. It relies on association, tracking and fusion algorithms to determine the true motion of the aircraft on both the vertical and horizontal axes, alongside contextual information useful for identifying the aircraft's flight mode at any given time. This paper focuses on the filtering of the vertical dimension and outlines certain problems present in the system: the effect of quantified measurements, the appearance of mode of flight transition overshoots and the low reactivity of the filter to abrupt transitions. These problems are analyzed and preliminary improvements according to the ATC context are implemented to overcome them. To demonstrate the improvement, a comparison between the proposed system and the original one is carried out through synthetic trajectory analysis.

ABSTRACT. One line of thinking is that measurement fusion at the system level provides enough value to warrant the cost of communicating the measurement level information to the system level. Another line of thinking is that in general the tracks generated at the source are sufficient to convey the information needed at the system level. There is no argument that measurements provide theoretically more information, but is the cost worth the possible extra value? This paper builds on our previous work that began to explore this issue with a straightforward simulation setup. In the previous paper it was shown that measurement fusion at the system level is better than just source track fusion at the system level. Those results were obtained when the update rate of measurements at the source were the same rate as the source tracks sent. In reality, the source tracks are produced at the full measurement rate, and the full measurement rate is not likely to be sent to the system level. Thus, this paper looks at disparate rates at the source level versus the system level. That is, the tracks produced at the source level will have a higher update rate than the rate at which measurements are sent to the system level to fuse.

ABSTRACT. This paper presents a multi-target metric driven approach to sensor management for Bernoulli filtering, in which at most one target of interest is present. The metric used is the generalised optimal sub pattern assignment (GOSPA) metric. We consider the problem of having an agile sensor operating in a surveillance area, tracking objects as they appear from a target birth distribution. Only one target of interest can exist at any given time-step and its single-target density is Gaussian. In this scenario, we have a grid of sensors that we can select from, one at a time using myopic planning. We evaluate the proposed sensor management algorithm via simulations.

ABSTRACT. This paper introduces several advances in methods to evaluate performance of multi-target trackers. First, we modify the classical target and track purity definitions to be more reflective of fragmentation and swap phenomena. Second, we extend performance evaluation to more general track outputs with arbitrary state probability distributions. Finally, we introduce a track-level generalization of the GOSPA metric that offers a scalar performance measure of overall tracking performance.

ABSTRACT. Random matrices are used to filter the center of gravity (CoG) and the covariance matrix of measurements. However, these quantities do not always correspond directly to the position and the extent of the object, e.g. when a lidar sensor is used. In this paper, we propose a Gaussian processes regression model (GPRM) to predict the position and extension of the object from the filtered CoG and covariance matrix of the measurements. Training data for the GPRM are generated by a sampling method and a virtual measurement model (VMM). The VMM is a function that generates artificial measurements using ray tracing and allows us to obtain the CoG and covariance matrix that any object would cause. This enables the GPRM to be trained without real data but still be applied to real data due to the precise modeling in the VMM. The results show an accurate extension estimation as long as the reality behaves like the modeling and e.g. lidar measurements only occur on the side facing the sensor.

ABSTRACT. In high-resolution automotive radar tracking systems, vehicle targets are often regarded as extended targets, which means multiple measurements originated from scattering centers of vehicle targets can be detected at each scan and thus the traditional point target tracking schemes are unsuitable. Meanwhile, vehicle maneuvers, e.g., braking and swerving, cause serious degradation of the classical extended target tracking methods. In this paper, a novel method is proposed for maneuvering extended vehicle tracking with automotive radar. The data-region association (DRA) strategy is adopted to handle the vehicle extension effect, which is superior in describing the complex spatial distribution of vehicle target measurements. The interacting multiple model (IMM) method is combined with this DRA strategy to describe the evolution of target motion models. Accordingly, the proposed DRA-IMM method achieves satisfying tracking performance of extended vehicles and also guarantees the robustness in case of maneuvers. Furthermore, in view of the correlation between vehicle extension and its kinematic state, a ray-based strategy is devised to improve the prior distribution of the data-region association of the basic DRA-IMM, and accordingly an enhanced DRA-IMM (EDRA-IMM) method is proposed. Simulation result validates the effectiveness of the proposed DRA-IMM method for maneuvering extended vehicle tracking and the further improvement of the proposed EDRA-IMM method.

ABSTRACT. In the past years, algorithms for 3D shape tracking using radial functions in spherical coordinates represented with different methods have been proposed. However, we have seen that mainly measurements from the lateral surface of the target can be expected in a lot of dynamic scenarios leading to an error-prone shape estimate in the top and bottom parts of the target when using a representation in spherical coordinates. We, therefore, propose to represent the shape of the target using a radial function in cylindrical coordinates, as this only represents regions of the lateral surface, and no information from the top or bottom parts is needed. In this paper, we use a Fourier-Chebyshev double series for 3D shape representation since a mixture of Fourier and Chebyshev series is a suitable basis for expanding a radial function in cylindrical coordinates. We investigated the method in a simulated and real-world maritime scenario with a CAD model of the target boat as a reference. We have found that shape representation in cylindrical coordinates has decisive advantages compared to a shape representation in spherical coordinates and should preferably be used if no prior knowledge of the measurement distribution on the surface of the target is available.

ABSTRACT. Weak target recognition, tracking and track management with a low signal-to-noise ratio (SNR) are always tricky problems. Probability hypothesis density (PHD) filtering propagates the first-order multi-target moment to obtain the best Poisson approximation to multi-target density. The PHD filtering does not consider explicit associations between measurements and targets, which is computationally efficient. But it cannot distinguish different targets or extract the time series of track states. Based on track-before-detect (TBD) strategies, this paper proposes labeled PHD (LPHD) filtering and derives its close-form solution, which identifies targets with a unique label. It is derived based on rigorous Bayes criteria, finite set statistics and Kullback-Leibler divergence minimization approximation. The separable TBD-based observation likelihood is conjugate to the Poisson mixture prior for LPHD filtering. Under the point-target assumption, the multi-hypothesis assignments of pixel-to-target are implemented with Murty's K-shortest path algorithm for LPHD filtering. Additionally, sequential Monte Carlo (SMC) implementations under the nonlinear non-Gaussian assumption are devised. Finally, simulations exhibit good performance in low SNR scenarios.

ABSTRACT. Passive location and tracking of radio emitters is of great research value in civilian and defense applications. Among the existing methods, localization based on received signal strength indicator (RSSI) has been widely used due to its advantages in terms of low cost and easy implementation. However, most RSSI-based localization methods rely on the assumption that the emitter has been detected. Moreover, the emitter signal is supposed to propagate with the simplified path-loss model in which the shadow effects caused by obstacles are not considered. As a result, there are still gaps between the aforementioned methods and practical applications. In this paper, we consider the combined path-loss and shadowing model, which has been empirically confirmed in both outdoor and indoor radio propagation environments. Joint detection and tracking of an emitter is proposed by modeling the state of the emitter as Bernoulli random finite set, characterized by an existence probability and a spatial probability density function. Compared to existing studies, this paper works upon more practically appealing signal propagation model, and achieves better performance in real-time emitter detection and tracking. Moreover, the proposed method also provides explicit estimates of the unknown shadowingrelated parameters, which can be adopted by further applications such as spectrum cartography and radio map construction. The feasibility of the proposed method is assessed via simulation experiments.

ABSTRACT. In object-tracking and state-estimation problems, ambiguous evidence such as imprecise measurements and the absence of detections can contain valuable information and thus be leveraged to further refine the probabilistic belief state. In particular, knowledge of a sensor’s bounded field of view (FoV) can be exploited to incorporate evidence of where an object was not observed. This paper presents a systematic approach for incorporating knowledge of the FoV geometry, position, and object inclusion/exclusion evidence into object state densities and random finite set multiobject cardinality distributions. The resulting state estimation problem is nonlinear and solved using a new Gaussian mixture approximation based on recursive component splitting. Based on this approximation, a novel Gaussian mixture Bernoulli filter for imprecise measurements is derived and demonstrated in a tracking problem using only natural language statements as inputs. This paper also considers the relationship between bounded FoVs and cardinality distributions for a representative selection of multiobject distributions, which can be used for sensor planning, as is demonstrated through a problem involving a multi-Bernoulli process with up to 100 potential objects.

ABSTRACT. The information provided by the vehicle’s sensors allows the estimation of critical parameters in a pre-crash scenario. However, the available open multimodal datasets focused on optical sensors usually present a low-resolution radar. Moreover, the sensors’ positioning does not provide a representative view of the objects in the last meters around them. With this in mind, we introduce a new approach to fuse camera, lidar, and radar of high resolution at a very early stage using a fully convolutional network and an anchor-free strategy to detect the front side of a vehicle, estimating distance and orientation in a single step. A robust experimental protocol based on a new multimodal dataset shows that the proposed fusion of the three sensors brings more stability and accuracy in detection and regression tasks by better representing far and near ranges around the vehicle where the sensors response to objects change significantly.

ABSTRACT. The recent fast advancements in the Internet of Video Things (IoVT) and Edge-Fog-Cloud Computing paradigm make smart public safety surveillance (SPSS) a realistic solution for an effective public safety service in smart cities. Typically, a fully functional SPSS system requires multiple sensory inputs for situational awareness (SAW). As an essential component in the context of highly complex, dynamic, and heterogeneous smart city operations, SPSS is expected to be environmentally resilient. Personal safety is among the top concerns of the residents in smart cities, and correspondingly pedestrian detectors are critical. Contemporary pedestrian detectors use optical cameras, whose accuracy is diminished in low-light environments, and they are rendered ineffective when obstacles block the direct line of sight to the camera. Thermal cameras are unaffected by such conditions and could be complementary. In this paper, we propose a hybrid pedestrian detection (HYPE) scheme based on the fusion of data obtained from the IoVT network consisting of optical and thermal cameras. We prove that training the YOLOv5 object detection model on a dataset of infrared images improves its accuracy in the detection of humans present in thermal surveillance images. We see a 33% decrease in object loss after transfer learning is performed.

ABSTRACT. Personalized monitoring systems (PMS) are used for Decision Making (DeM) to support humans and fully autonomous decision-making in several applications, such as health monitoring and management. However, it is still challenging to design decision-making in PMS. In this work, we propose a systematic modeling approach, called DEMDE, for decision-making design in PMS during the design phase. DEMDE guides the design of a specific Bayesian network (BN) from an instantiated domain model for PMS based on context-aware data fusion using a probabilistic domain model (general BN). We evaluated our proposal by developing a BN for decision-making about sending an alert of high contamination risk in cell culture. The case study demonstrated the application of the DEMDE concepts and the modeling process, including model evaluation through sensitivity analysis to assess the robustness and reliability of the modeled BN.

ABSTRACT. Multi-sensor fusion can improve environment perception, e.g., by increasing the field of view in collective perception, where intelligent vehicles communicate. Track-to-track fusion in a collective perception scenario makes it necessary to associate tracks from multiple sensors. Especially in settings with many sensors that have limited field of view, track-to-track association can be quite challenging. In this work, we develop a stochastic optimization-based approach for an arbitrary number of sensors with a limited field of view, which utilizes a cluster likelihood to sample joint associations. The proposed method clearly outperforms a greedy approach and computes the most likely associations with only a few samples. We evaluate the approaches on simulated data in static and dynamic scenarios.

ABSTRACT. Network-centric multitarget tracking under communication constraints is considered, where dimension-reduced track estimates are exchanged. Previous work on target tracking in this subfield has focused on fusion aspects only and derived optimal ways of reducing dimensionality based on fusion performance. In this work we propose a novel problem formalization where estimates are reduced based on association performance. The problem is analyzed theoretically and problem properties are derived. The theoretical analysis leads to an optimization strategy that can be used to partly preserve association quality when reducing the dimensionality of communicated estimates. The applicability of the suggested optimization strategy is demonstrated numerically in a multitarget scenario.

ABSTRACT. —In this work a new class of filters, called Lambda:Omicron Multiplicative Error Model (L:OMEM), is introduced with the aim to solve efficiently the tracking problem of maneuvering extended objects. In this context two main challenges have to be solved: (1) the tracked object moves with an unknown time-varying speed and an unknown turning rate; (2) the tracked object can generates a large amount of measurements. Closed-form formulas and a novel method to reduce the extended object tracking problem to a conventional point object tracking problem are derived, so that the novel filter results in an accurate and an extremely low computational cost algorithm. Numerical simulations are presented to validate the effectiveness of the proposed approach, where the L:OMEM filter is compared against state of the art filters for extended objects.

ABSTRACT. This paper addresses the joint detection and tracking of an unknown and time-varying number of targets in clutter. Here we formulate the tracking task in a variable-dimension state space, under which the reversible jump sequential Markov chain Monte Carlo sampling methods can be utilised to online estimate the target number, their kinematic states, and the association variables. In particular, a novel Rao-Blackwellisation scheme is devised to improve the tracking accuracy and sampling efficiency for linear Gaussian models. Based on the non-homogeneous Poisson process measurement model, the developed tracker enjoys a partially parallel sampling structure, thereby being able to efficiently tackle the data association under massive measurements and clutter. The simulation results demonstrate that the developed tracker exhibits superior tracking performance in comparison to existing trackers in both accuracy and computational efficiency when tracking multiple targets under heavy clutter.

ABSTRACT. We address the magnetic-field simulteanous localization and mapping (SLAM) problem for global positioning. We leverage a previously-developed particle filter (PF)-based framework for online Bayesian inference and learning of Gaussian Process state-space models (GP-SSMs). We extend the framework to directly incorporate physical properties of the magnetic field in the GP formulation, by leveraging that magnetic fields under the absence of free currents are curl free. Because of its flexibility, the method can include any motion model that can be expressed by a general nonlinear function, with potential applications to, e.g., mobile robotics and pedestrian localization. Simulation results indicate that our method performs similar to recent batch methods for magnetic-field slam, while the computation times are feasible for online implementations.

ABSTRACT. Practical multi-hypothesis trackers (MHTs) often entail a number of parameters for track confirmation and extraction logic, gating and pruning, most of which are chosen heuristically to tradeoff performance and computational cost. Conceptually, these parameters are unnecessary with optimal MHT processing, as these decisions will fall out from the optimal solution, though perhaps with an increase in processing cost. We demonstrate, however, with a canonical MHT model and its attendant association assignment problem that many of these parameters can be chosen losslessly, that is, they only remove hypotheses that an optimal association solution is guaranteed to remove anyway, and thus strictly improve computational cost, with no loss in tracking performance. At the other end of the spectrum, the tools developed likewise yield a number of relations that detect when parameters are set such that practical tracking is no longer possible.

ABSTRACT. The proliferation of large satellite constellations in low Earth orbit (LEO) is dramatically increasing demand on existing systems for space domain awareness. The rapidly growing number of objects in LEO will reduce the average rate of observations per object, necessitating the development of multi-target algorithms that can handle higher levels of data sparsity without sacrificing computational efficiency. In this paper, we demonstrate that a multi-target filter that combines a number of useful capabilities is able to track and maintain custody of a simulated population of over 16,000 LEO objects without requiring the use of high-performance computing facilities. The filter is based on the generalized labeled multi-Bernoulli filter, and includes three previously-presented features: label space partitioning based on sensor fields of view, the ensemble Gaussian mixture filter (EnGMF), and bi-fidelity orbit uncertainty propagation. We also introduce a new, algorithmically simple method for adjusting the number of EnGMF particles to balance accuracy and computational efficiency, which we refer to as progressive resampling.

ABSTRACT. Multi-Target Tracking (MTT) is a challenging problem in the field of data association and sensor data fusion. Many solutions to MTT assume a Markovian nature to the motion of the target to solve the problem and avoid the potential computational complexity. Recently, we have shown that considering a sequence of three time steps and their resulting triplet costs in data association provides a superior solution that better incorporates the kinematic behavior of maneuvering targets. Nevertheless, the triplet costs pose significant computational overhead and scaling challenges. In this paper, we present significant computational advances in a triplet cost-based data association engine for MTT using Graphics Processing Units (GPUs). We achieve this by improving the computational performance of the dual ascent algorithm for dense Multi-Dimensional Assignment Problem (MAP), presented in Vadrevu and Nagi, 2022. Our contributions include: (1) A very fast GPU-accelerated Linear Assignment Problem (LAP) solver that solves an array of tiled LAPs without synchronizing with the CPU, (2) Reduction in computational overheads of triplet costs by using gating and compressed matrix representations, and (3) Computational performance studies that demonstrate the effectiveness of our computational enhancements. Our resulting solution is 5.8 times faster than the current solution (Vadrevu and Nagi, 2022) without compromising the accuracy.