bat365 Homepage

Multi-Object Trackers

Multi-sensor multi-object trackers, data association, and track fusion

You can create multi-object trackers that fuse information from various sensors. Use trackerGNN to maintain a single hypothesis about the tracked objects. Use trackerTOMHT to maintain multiple hypotheses about the tracked objects. Use trackerJPDA to assign multiple probable detections to the tracked objects. Use trackerPHD to represent tracked objects using probability hypothesis density (PHD) function. Use trackerGridRFS to track objects using a grid-based occupancy evidence approach. Use trackFuser to fuse tracks generated by tracking sensors or trackers and architect decentralized tracking systems.

Functions

expand all

Assignment

`assignauction`	Assignment using auction global nearest neighbor
`assignjv`	Jonker-Volgenant global nearest neighbor assignment algorithm
`assignkbest`	Assignment using k-best global nearest neighbor
`assignkbestsd`	K-best S-D solution that minimizes total cost of assignment
`assignmunkres`	Munkres global nearest neighbor assignment algorithm
`assignsd`	S-D assignment using Lagrangian relaxation
`assignTOMHT`	Track-oriented multi-hypotheses tracking assignment
`jpdaEvents`	Feasible joint events for trackerJPDA (Since R2019a)
`partitionDetections`	Partition detections based on distance (Since R2019a)
`mergeDetections`	Merge detections into clustered detections (Since R2021b)

Trackers

`trackerGNN`	Multi-sensor, multi-object tracker using GNN assignment
`trackerJPDA`	Joint probabilistic data association tracker (Since R2019a)
`trackerTOMHT`	Multi-hypothesis, multi-sensor, multi-object tracker
`trackerPHD`	Multi-sensor, multi-object PHD tracker (Since R2019a)
`trackerGridRFS`	Grid-based multi-object tracker (Since R2020b)
`smootherJIPDA`	Joint probabilistic data association smoother (Since R2023a)
`dynamicEvidentialGridMap`	Dynamic grid map output from `trackerGridRFS` (Since R2021a)
`objectDetection`	Report for single object detection
`objectDetectionDelay`	Simulate out-of-sequence object detections (Since R2022a)
`getTrackPositions`	Returns updated track positions and position covariance matrix
`getTrackVelocities`	Obtain updated track velocities and velocity covariance matrix
`clusterTrackBranches`	Cluster track-oriented multi-hypothesis history
`compatibleTrackBranches`	Formulate global hypotheses from clusters
`pruneTrackBranches`	Prune track branches with low likelihood
`trackHistoryLogic`	Confirm and delete tracks based on recent track history
`trackScoreLogic`	Confirm and delete tracks based on track score
`trackBranchHistory`	Track-oriented MHT branching and branch history
`trackingSensorConfiguration`	Represent sensor configuration for tracking (Since R2019a)

Track and Detection Fusion

`trackFuser`	Single-hypothesis track-to-track fuser (Since R2019b)
`trackingArchitecture`	Tracking system-of-system architecture (Since R2021a)
`staticDetectionFuser`	Static fusion of synchronous sensor detections
`objectTrack`	Single object track report (Since R2019b)
`fusecovint`	Covariance fusion using covariance intersection
`fusecovunion`	Covariance fusion using covariance union
`fusexcov`	Covariance fusion using cross-covariance
`fuserSourceConfiguration`	Configuration of source used with track fuser (Since R2019b)
`triangulateLOS`	Triangulate multiple line-of-sight detections

Blocks

expand all

Trackers

Global Nearest Neighbor Multi Object Tracker	Multi-sensor, multi-object tracker using GNN assignment (Since R2019b)
Joint Probabilistic Data Association Multi Object Tracker	Joint probabilistic data association tracker (Since R2019b)
Track-Oriented Multi-Hypothesis Tracker	Track-Oriented Multi-Hypothesis Tracker (Since R2020a)
Probability Hypothesis Density (PHD) Tracker	Multi-sensor, multi-object PHD tracker (Since R2021a)
Grid-Based Multi Object Tracker	Grid-based multi-object tracker using random finite set approach (Since R2021b)
Track-To-Track Fuser	Track-to-Track Fusion (Since R2021a)
Detection Concatenation	Combine detection reports from different sensors (Since R2021a)
Track Concatenation	Concatenate tracks (Since R2021a)

Topics

Introduction to Multiple Target Tracking
Introduction to assignment-based multiple target trackers.
Introduction to Assignment Methods in Tracking Systems
Introduce 2-D and S-D assignment problems in tracking systems.
Introduction to Track-To-Track Fusion
Track-To-Track Fusion Architecture Using Track Fuser.
Multiple Extended Object Tracking
Introduction to methods and examples of multiple extended object tracking in the toolbox.
Convert Detections to objectDetection Format
These examples show how to convert actual detections in the native format of the sensor into objectDetection objects.
Introduction to Using the Global Nearest Neighbor Tracker
This example shows how to configure and use the global nearest neighbor (GNN) tracker.
Introduction to Track Logic
This example shows how to define and use confirmation and deletion logic that are based on history or score.
Introduction to PHD Filter
This example introduces the principles behind the probability hypothesis density (PHD) filter and how it can be used to estimate the number and states of multiple objects in a scene.
Generate Code with Strict Single-Precision and Non-Dynamic Memory Allocation
Introduce functions, objects, and blocks that support strict single-precision and non-dynamic memory allocation code generation in Sensor Fusion and Tracking Toolbox™.

Featured Examples

Tracking Closely Spaced Targets Under Ambiguity

Track objects when the association of sensor detections to tracks is ambiguous. In this example, you use a single-hypothesis tracker, a multiple-hypothesis tracker, and a probabilistic data association tracker to compare how the trackers handle this ambiguity. To track, the maneuvering targets better, you estimate the motion of the targets by using various models.

Open Script

Introduction to Class Fusion and Classification-Aided Tracking

Perform class fusion using multi-object tracker.

Open Live Script

Tracking a Flock of Birds

Track a large number of objects. A large flock of birds is generated and a global nearest neighbor multi-object tracker, trackerGNN, is used to estimate the motion of every bird in the flock.

Open Script

Define and Test Tracking Architectures for System-of-Systems

Define the tracking architecture of a system-of-systems that includes multiple detection-level multi-object trackers and track-level fusion algorithms. You can use the tracking architectures to compare different tracking system designs and find the best solution for your system.

Open Live Script

How to Efficiently Track Large Numbers of Objects

Use the trackerGNN to track large numbers of targets. Similar techniques can be applied to the trackerJPDA and trackerTOMHT as well.

Open Script

Tuning a Multi-Object Tracker

Tune and run a tracker to track multiple objects in the scene. The example explains and demonstrates the importance of key properties of the trackers in the Sensor Fusion and Tracking Toolbox.

Open Live Script

Analyze Track and Detection Association Using Analysis Info

Use tracker analysis information to analyze detection and track association in tracking process.

Open Live Script

Track Multiple Lane Boundaries with a Global Nearest Neighbor Tracker

Design and test a multiple lane tracking algorithm. The algorithm is tested in a driving scenario with probabilistic lane detections.

Open Live Script

Handle Out-of-Sequence Measurements in Multisensor Tracking Systems

Handle out-of-sequence measurements (OOSM) in a multisensor tracking system. The example compares tracking results when OOSM are present using various handling techniques. For more information about OOSM handling techniques see Handle Out-of-Sequence Measurements with Filter Retrodiction example.

Open Live Script

Introduction to JIPDA Smoothing

Introduce joint integrated probabilistic data association multi-object smoothing algorithm and its applications.

Open Live Script

Understand and Analyze JIPDA Smoother Algorithm

Illustrate workflow of joint integrated probabilistic data association multi-object smoother algorithm.

Open Live Script

Define and Test Tracking Architectures for System-of-Systems in Simulink

Define architectures for a tracking system-of-systems in MATLAB and export them to a Simulink model. You compare various tracking system designs that includes multiple detection-level multi-object trackers and track fusers in Simulink. You use Simulink Variant systems to realize different architecture solutions for your system. This example closely follows the Define and Test Tracking Architectures for System-of-Systems MATLAB example.

Open Live Script

Export trackingArchitecture to Simulink

Define a trackingArchitecture object and export it to Simulink.

Open Live Script

Track Simulated Vehicles Using GNN and JPDA Trackers in Simulink

Configure and utilize GNN and JPDA trackers in a simulated highway scenario in Simulink® with Sensor Fusion and Tracking Toolbox™. It closely follows the Sensor Fusion Using Synthetic Radar and Vision Data in Simulink (Automated Driving Toolbox). A main benefit of modeling the system in Simulink is the simplicity of performing "what-if" analysis and choosing a tracker that results in the best performance based on the requirements.

Open Model

Track Closely Spaced Targets Under Ambiguity in Simulink

Track objects in Simulink® with Sensor Fusion and Tracking Toolbox™ when the association of sensor detections to tracks is ambiguous. It closely follows the Tracking Closely Spaced Targets Under Ambiguity MATLAB® example.

Open Model

How to Generate C Code for a Tracker

Generate C code for a MATLAB function that processes detections and outputs tracks. The function contains a trackerGNN, but any tracker can be used instead.

Open Script

Generate Code for a Track Fuser with Heterogeneous Source Tracks

Generate code for a track-level fusion algorithm in a scenario where the tracks originate from heterogeneous sources with different state definitions. This example is based on the Track-Level Fusion of Radar and Lidar Data example, in which the state spaces of the tracks generated from lidar and radar sources are different.

Open Live Script