Skip to Content
This documentation is provided with the HEAT environment and is relevant for this HEAT instance only.
RunnersAviation MetricsAviation Metrics Runner

Aviation Metrics Runner

The Aviation Metrics Runner processes biometric, simulation, and eye-tracking data collected during aviation training sessions. It produces scored outputs covering cognitive load, lookout scan behaviour, flight-path adherence, circuit performance, and more. These outputs feed downstream dashboard transforms that render instructor-facing dashboards.

Note: Several node templates in this runner (notably cognitive-load-metrics) are domain-agnostic and may be extracted into their own runner in a future release.


Available Node Templates

1. am-preprocess (Transform Node)

Pre-processes raw biometric and simulation data into the aligned format expected by downstream analytics nodes.

Typical usage: First node in a legacy aviation pipeline; aligns timestamps and normalises column names.

2. rapid-preprocess (Transform Node)

Rapid-format pre-processing that merges biometric and simulation streams into a single JSON payload containing bio_input and sim_input arrays.

Typical usage: Entry point for RAPID MACE pipelines. Downstream nodes such as cognitive-load-metrics, lookout-scan-metrics, and phase-segment consume its output.

3. phase-segment (Transform Node)

Segments a session into named flight phases (e.g. take-off, upwind, downwind, final approach) based on simulation telemetry.

Typical usage: Provides per-phase slices to cognitive-load-metrics, lookout-scan-metrics, and other analytics nodes.

4. cognitive-load-metrics (Processing Node)

Computes a cognitive load score, timeline, and zone breakdown from biometric data using an XGBoost regression model over sliding windows of eye-tracking and heart-rate features. Accepts optional smoothing and classification configuration.

Typical usage: Placed after rapid-preprocess (and optionally phase-segment) to produce cognitive load outputs consumed by timeline and dashboard transforms. See the cognitive-load-metrics guide.

5. lookout-scan-metrics (Processing Node)

Analyses gaze data to score lookout scan behaviour against expected scan patterns for the current flight phase.

6. timeline-metrics (Processing Node)

Aggregates multiple upstream metric timelines (cognitive load, lookout scan, etc.) into a unified timeline structure for dashboard rendering.

7. weather-overview-metrics (Processing Node)

Extracts and summarises weather conditions from simulation telemetry for contextual display on dashboards.

8. gaze-eye-metrics (Processing Node)

Classifies 3D gaze direction data into spatial zones (Far, Near, Left, Right) using KMeans clustering with Mahalanobis distance augmentation. Produces overall gaze percentages, raw gaze vectors, and a per-second timeline. Single-user scoped.

Typical usage: Placed after rapid-preprocess (and optionally phase-segment) to produce gaze distribution outputs consumed by dashboard transforms. See the gaze-eye-metrics guide.

9. flight-path-metrics (Processing Node)

Evaluates flight-path adherence by comparing actual trajectory data against reference profiles.

10. circuit-scoring (Processing Node)

Scores individual flying circuit segments (take-off, upwind, downwind, final turn, final approach) against configurable criteria.

11. am-transform-circuits-dashboard (Transform Node)

Transforms scored analytics outputs into the schema expected by the circuits dashboard.


Example Workflow