A Multisensor Multi-Bernoulli Filter

Saucan, Augustin-Alexandru; Coates, Mark; Rabbat, Michael

doi:10.1109/tsp.2017.2723348

Cited by 84 publications

(65 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Random finite sets, generalized labeled multi-Bernoulli, multi-object tracking, data association, Gibbs sampling The classical PHD and CPHD filters are developed for single-sensors. Since the multi-sensor PHD, CPHD and multi-Bernoulli filters are combinatiorial [4], [30], the most commonly used approximate multi-sensor PHD, CPHD and multi-Bernoulli filter are the heuristic "iterated corrector" versions [31] that apply single-sensor updates, once for each sensor in turn. This approach yields final solutions that depend on the order in which the sensors are processed.…”

Section: Index Termsmentioning

confidence: 99%

See 1 more Smart Citation

Multi-Sensor Multi-Object Tracking With the Generalized Labeled Multi-Bernoulli Filter

Beard

2019

IEEE Trans. Signal Process.

126

133

View full text Add to dashboard Cite

This paper proposes an efficient implementation of the multi-sensor generalized labeled multi-Bernoulli (GLMB) filter. The solution exploits the GLMB joint prediction and update together with a new technique for truncating the GLMB filtering density based on Gibbs sampling. The resulting algorithm has quadratic complexity in the number of hypothesized object and linear in the number of measurements of each individual sensors. Index TermsRandom finite sets, generalized labeled multi-Bernoulli, multi-object tracking, data association, Gibbs sampling The classical PHD and CPHD filters are developed for single-sensors. Since the multi-sensor PHD, CPHD and multi-Bernoulli filters are combinatiorial [4], [30], the most commonly used approximate multi-sensor PHD, CPHD and multi-Bernoulli filter are the heuristic "iterated corrector" versions [31] that apply single-sensor updates, once for each sensor in turn. This approach yields final solutions that depend on the order in which the sensors are processed. Multi-sensor PHD and CPHD filters that are principled, computationally tractable, and independent of sensor order have been proposed in [4] (Section 10.6).However, this approach as well as the heuristic "iterated corrector" involve two levels of approximation since the exact multi-sensor PHD, CPHD and multi-Bernoulli filters are approximations of the Bayes multi-sensor multi-object filter.An exact solution to the Bayes multi-object filter is the Generalized Labeled Multi-Bernoulli (GLMB) filter, which also outputs multi-object trajectories [32], [33]. Moreover, given a cap on the number of GLMB components, recent works show that the GLMB filter can be implemented with linear complexity in the number of measurements and quadratic in the number of hypothesized objects [34]. The GLMB density is flexible enough to approximate any labeled RFS density with matching intensity function and cardinality distribution [35], and also enjoys a number of nice analytical properties, e.g. the void probability functional-a necessary and sufficient statistic-of a GLMB, the Cauchy-Schwarz divergence between two GLMBs, the L 1 -distance between a GLMB and its truncation, can all be computed in closed form [36], [33]. Recent research in approximate GLMB filters [37], [38] as well as applications in tracking from merged measurements [39], extended targets [40], maneuvering targets [41], [42], trackbefore-detect [35], [43], computer vision [44]-[47], sensor scheduling [36], [48], field robotics [49], and distributed multi-object tracking [50], demonstrate the versatility of the GLMB filter, and suggest that it is an important tool in multi-object systems.In this work we present an implementation of the multi-sensor GLMB filter. The major hurdle in the multi-sensor GLMB filter implementation is the NP-hard multi-dimensional ranked assignment problem.A multi-sensor version of an approximation of the GLMB filter, known as the marginalized GLMB filter, was proposed in [38]. While this multi-sensor solution is scalable in the number of sensors, it...

show abstract

Section: Index Termsmentioning

confidence: 99%

“…Instead of sampling from π(I + , θ + |I, ξ) ∝ ω (I,ξ,I+,θ+) Z+ , this subsection introduces an alternative target distribution for the Gibbs sampler, which can drastically reduce the complexity. Unique samples from the alternative target distribution are then reweighted according to (30).…”

Section: B Alternative Target Distributionmentioning

confidence: 99%

Multi-Sensor Multi-Object Tracking With the Generalized Labeled Multi-Bernoulli Filter

Beard

2019

IEEE Trans. Signal Process.

126

133

View full text Add to dashboard Cite

show abstract

“…However, it has been reported that the estimation results depend on the order in which the RDs are processed [36]. An alternative way to mitigate the computational load is to directly reduce the number of partitions and associations in each partition [33], [34], [37]- [39]. In [37], a distance partition method is proposed for extended target tracking.…”

Section: Introductionmentioning

confidence: 99%

“…Then, by applying the l max partition to the partitioned measurement set, the number of resulting partitions can be further reduced. In [38], [39], a greedy partition method is proposed to reduce the number of partitions. It considers only a certain number of partitions with the highest weight at every step and discards the others.…”

Section: Introductionmentioning

confidence: 99%

TOA-Based Indoor Localization and Tracking With Inaccurate Floor Plan Map via MRMSC-PHD Filter

2019

View full text Add to dashboard Cite

This paper proposes a novel indoor localization scheme to jointly track a mobile device (MD) and update an inaccurate floor plan map using the time-of-arrival measured at multiple reference devices (RDs). By modeling the floor plan map as a collection of map features, the map and MD position can be jointly estimated via a multi-RD single-cluster probability hypothesis density (MSC-PHD) filter. Conventional MSC-PHD filters assume that each map feature generates at most one measurement for each RD. If single reflections of the detected signal are considered as measurements generated by map features, then higher-order reflections, which also carry information on the MD and map features, must be treated as clutter. The proposed scheme incorporates multiple reflections by treating them as virtual single reflections reflected from inaccurate map features and traces them to the corresponding virtual RDs (VRDs), referred to as a multi-reflection-incorporating MSC-PHD (MRMSC-PHD) filter. The complexity of using multiple reflection paths arises from the inaccuracy of the VRD location due to inaccuracy in the map features. Numerical results show that these multiple reflection paths can be modeled statistically as a Gaussian distribution. A computationally tractable implementation combining a new greedy partitioning scheme and a particle-Gaussian mixture filter is presented. A novel mapping error metric is then proposed to evaluate the estimated map?s accuracy for plane surfaces. Simulation and experimental results show that our proposed MRMSC-PHD filter outperforms existing MSC-PHD filters by up to 95% in terms of average localization and by up to 90% in terms of mapping accuracy.

show abstract

“…And to improve the estimation accuracy, the multiBernoulli smoother which consists of original CBMeMBer filter and backward smoothing was proposed in [11]. In addition, the multi-Bernoulli filter was also enhanced for multi-sensor tracking [12,13], extended target tracking [14,15], multipath multi-target tracking [16], and group object tracking [17].…”

Section: Introductionmentioning

confidence: 99%

Measurement-Driven Multi-Target Multi-Bernoulli Filter

Lei

2018

Mathematical Problems in Engineering

View full text Add to dashboard Cite

A measurement-driven multi-target multi-Bernoulli (MeMBer) filter which modifies the MeMBer filter by the measurements information is proposed in this paper. The proposed filter refines both the legacy estimates and the data-induced estimates of the MeMBer filter. For the targets under the legacy track set, the detection probabilities derived from the measurements are employed to refine the multi-target distribution. And for the targets under the data-induced track set, the multi-target distribution is further improved by the modified existence probabilities of the legacy tracks. Unlike the cardinality balanced MeMBer (CBMeMBer) filter, the proposed filter removes the cardinality bias in the MeMBer filter by utilizing the measurements information. Simulation results show that, compared with the traditional methods, the proposed filter can improve the stability and accuracy of the estimates and does not need the high detection probability hypothesis.

show abstract

A Multisensor Multi-Bernoulli Filter

Cited by 84 publications

References 19 publications

Multi-Sensor Multi-Object Tracking With the Generalized Labeled Multi-Bernoulli Filter

Multi-Sensor Multi-Object Tracking With the Generalized Labeled Multi-Bernoulli Filter

TOA-Based Indoor Localization and Tracking With Inaccurate Floor Plan Map via MRMSC-PHD Filter

Measurement-Driven Multi-Target Multi-Bernoulli Filter

Contact Info

Product

Resources

About