Create a track
Publish fusion-ready tracks from sensors so Lattice can correlate them
This page shows how to publish tracks that the Lattice fusion engine can ingest and correlate. It covers 2D bearing-only tracks, 3D signal-of-interest tracks with an estimated range, and fully resolved 3D tracks with position and velocity uncertainty. For more information on entities, see Entities overview.
A track represents any entity tracked by another asset or integration source. Tracks are not directly under the control of friendly forces. This includes aircraft tracks from radar or sensor hits, signal detections, and vehicles, people, or animals detected through cameras.
Before you begin
-
To publish tracks, set up the Lattice SDK and set up a Lattice environment.
-
Publish the observing sensor as a
TEMPLATE_ASSETentity before generating tracks. The fusion engine uses the sensor’s entity ID, which you specify in each track’srelationships.trackedBy, to identify which sensors contribute to the track. The examples on this page pair with the sensors from Create an asset: -
Choose a unique
integration_nameanddata_typepair. The fusion engine uses the combination to distinguish your track feed from other sources.
These gRPC examples authenticate with a static environment token attached as request metadata. If you are using OAuth 2.0 client credentials instead, set up the token refresh module before running the examples on this page.
Produce fusion-ready tracks
To make a track eligible for correlation by the Lattice fusion engine, populate the following components:
Generate with PublishEntity
Call PublishEntity, or
PublishEntities (gRPC-only)
to publish multiple entities in a single request.
Set identity and lifetime fields
Stable identifier for the track. Use the same value on every update so the fusion engine can correlate successive observations into a single track.
Human-readable label displayed in the Lattice common operational picture.
Time the track was first observed. Set this once and do not change it on subsequent updates; the fusion engine uses it to establish track age.
Future time at which Lattice automatically removes your contribution to the track. Set to approximately 10 seconds ahead of the current time on every update. While your sensor keeps detecting the target, each new detection refreshes the expiry and keeps your contribution live. When your sensor stops detecting, your contribution expires; the fused track persists as long as any other sensor continues to contribute. For more information, see Manage update rates.
Additional identifiers from the source system, such as ALT_ID_TYPE_CALLSIGN,
ALT_ID_TYPE_LINK16_TRACK_NUMBER, ALT_ID_TYPE_VMF_URN, or ALT_ID_TYPE_SERIAL_NUMBER. Populate one entry per identifier so
downstream systems can cross-reference tracks.
Streaming endpoints reject no_expiry. Always set a finite expiry_time when you
publish over a stream using PublishEntities.
Populate position and uncertainty
Resolved position of the track, including latitude, longitude, and HAE altitude. Populate this field for all 3D tracks.
East-North-Up (ENU) position covariance matrix expressing position uncertainty. A diagonal-only matrix is valid when the full matrix is unavailable.
Link-16 track quality score from 0 to 15. Use this instead of position_enu_cov only when
your source system produces a track-quality score rather than a covariance matrix.
Set ontology
Classification template that tells Lattice how to render and handle the entity. Set to
TEMPLATE_TRACK for resolved positional tracks, or TEMPLATE_SIGNAL_OF_INTEREST for RF
emitter tracks and 2D camera sensor tracks.
Set provenance and update cadence
Name of the integration publishing the track. Together with data_type, this pair uniquely
identifies your feed so the fusion engine can distinguish it from other sources.
Type of data being published, such as camera_lob or rf_sigint. Together with
integration_name, this pair uniquely identifies your feed.
Timestamp of the measurement that produced this update. Set it to the time your sensor
observed the target. Lattice uses it to detect stale tracks and rejects any update whose
source_update_time is earlier than or equal to the value it already holds. Publish a new
update only when your sensor produces a new detection.
Declare contributing sensors
List of sensor relationships for the track. Populate one entry per contributing sensor so the fusion engine knows which sensors produced the observation.
Entity ID of the sensor asset that produced this track. Publish the sensor asset before any track that references it.
Stable, unique identifier for this sensor-to-track relationship. A common pattern is
<track_id>-tracked-by-<sensor_id>.
Describes the sensor system actively tracking the entity. Include
actively_tracking_sensors with sensor_id, sensor_type, and operational_state values
that match the corresponding entries in the tracking entity’s top-level sensors component.
Set last_measurement_timestamp to the time of the most recent measurement.
For single-sensor tracks, publish one trackedBy relationship. For tracks detected by
multiple sensors in your system, publish one trackedBy relationship per contributing
sensor that exists in Lattice as an asset entity.
Choose a line-of-bearing component by sensor modality
A single track must carry its line-of-bearing data in exactly one component, selected by the sensing modality:
Line-of-bearing measurement for radio-frequency sensors. Use this component instead of
tracked.line_of_bearing when the observing sensor is an RF direction-finding system.
Line-of-bearing measurement for camera and acoustic sensors. Use this component instead of
signal.line_of_bearing when the observing sensor is a camera or acoustic array.
Bearing and elevation covariance in radians squared. Populate this field on whichever line-of-bearing component you use; the fusion engine uses it to decide whether to fuse the bearing or render it visually only.
Improve fusion quality with optional fields
Measured radar cross-section of the target. Optional; the fusion engine uses it to improve track association when the value is present.
Estimated length of the target in meters. Optional; the fusion engine uses it alongside radar cross-section to improve association accuracy when the value is present.
Symbology
Populate the following components on every track so downstream rendering and handling are correct:
Nationality of the track, used downstream to pick the flag and the MIL-STD-2525C country code.
Set explicitly when known; leaving it as NATIONALITY_INVALID produces generic symbology.
Valid 15-character MIL-STD-2525C Symbol Identification Code. Required for downstream command-and-control systems such as NGC2 to render the track; entities without a SIDC fall back to a generic icon.
Flags that describe the operational context of the track. Set simulated for all tracks
published during test and evaluation exercises. Additional flags, including exercise,
emergency, and c2, are available when applicable.
Classification level of the track data. Required when publishing classified content so Lattice and downstream systems can handle the data according to its classification.
MIL-STD-2525D is not supported by the current schema. Use 2525C and populate
symbology.mil_std_2525_c.sidc.
Publish a 2D bearing-only track
Use a 2D bearing-only track when your sensor, such as a camera or acoustic array, reports a
direction to the target but not a range. Populate
tracked.line_of_bearing
with the sensor’s position at the time of the measurement and the angle pointing toward the target.
Define the entity model’s required components. To represent a bearing-only detection:
- Set
templatetoTEMPLATE_SIGNAL_OF_INTEREST. - Populate
tracked.lineOfBearing.angleOfArrivalwith the sensor’s pose and bearing/elevation covariance. - Add a
trackedByrelationship that references the sensor’s entity ID.
Publish a 3D track
Use a 3D track when your sensor can place the target in three-dimensional space. Choose the variant that matches the measurement your sensor produces.
RF signal with line-of-bearing
Resolved 3D detection
Use this approach for a signal-of-interest track that has a bearing and an estimated range to
the emitter, such as a track produced by an RF direction-finding sensor. Set the bearing and range
in the
signal component rather than
the tracked component.
Define the entity model’s required components. To represent an RF track with bearing and range:
- Set
templatetoTEMPLATE_SIGNAL_OF_INTEREST. - Populate
signal.lineOfBearing.angleOfArrivalwith the sensor’s pose. - Populate
signal.lineOfBearing.rangeEstimateMwith the estimated distance and one-sigma uncertainty.
To report a track’s heading, populate location.attitude_enu using the same conventions as assets.
For more information, see Orientation conventions.
Manage update rates
Your feed is one of many contributing to a track. Lattice fuses the independent observations from every contributing sensor into a single track, and the accuracy of that fused picture depends on each feed reporting only what its own sensor actually observes. Your sensor is not solely responsible for keeping a track alive: when your sensor stops detecting the target, other sensors in the environment may still be detecting it, and the fusion engine continues to maintain the track from their contributions.
Publish a new update only when your sensor produces a new kinematic detection. Each update
must carry a provenance.source_update_time that reflects the time of that detection. Lattice
rejects any update whose source_update_time is earlier than or equal to the value it already
holds, so do not republish unchanged data to keep a track alive.
Don’t republish stale detections to keep a track alive. Republishing an unchanged detection tells the fusion engine your sensor is still observing the target when it isn’t, which degrades the aggregate picture that every other sensor contributes to. For example, republishing the same kinematic state in the absence of detections makes an air vehicle appear to hover.
Track liveness is governed by expiry_time. Set expiry_time to approximately 10
seconds ahead of the current time on every update. While your sensor keeps detecting the
target, each new detection refreshes the expiry and your contribution stays live. When your
sensor stops detecting, your contribution expires, and the fused track persists as long as any
other sensor continues to contribute.
The preceding examples loop to publish successive detections from a sensor, updating
source_update_time and expiry_time on each iteration. In a real integration, drive each
iteration from a new sensor measurement rather than a fixed timer.
What’s next?
- Follow your published tracks in real time with Cue and monitor.
- Review the full entity model in the Entities overview.
- Explore every field in the
PublishEntityAPI reference.