Signal K Dead Reckoning

An offline-first dead reckoning and sensor fusion engine for Signal K that maintains a continuously computed "shadow boat" position from water-track sensors (speed through water, compass heading, and learned leeway and current corrections), so you always have a navigational fallback when GPS becomes unreliable — whether from jamming, spoofing, or plain receiver failure. While GPS is trusted, the engine learns vessel-specific calibration corrections against ground truth and watches for GPS anomalies; when it isn't, the same learned model keeps the dead-reckoned position, its uncertainty polygon, and a water-track log going. Fixes from celestial sights, compass bearings, and vertical angles are entered through a unified pipeline and can snap dead reckoning back on track, with optional write-through to signalk-logbook.

Note: This is just a toy. Make your own navigation calculations and decisions.

Requirements

• This plugin installed and configured
• Some chart source available to Signal K
• Hand bearing compass and/or sextant

Getting started

1. Install the plugin from the Signal K Appstore (or npm install @meri-imperiumi/signalk-dead-reckoning) and enable it. The defaults are sane — you only need to revisit the configuration if you want to change intervals or integrations.
2. Make sure the required data feeds are available on the vessel network:
   • navigation.position (GPS) — the training baseline and anomaly reference
   • navigation.speedThroughWater (paddlewheel log) and navigation.headingMagnetic (compass + variation) — the water-track inputs
   • Wind data — used by the calibration model and the polar speed fallback
3. Just sail. While GPS is reliable and you're under sail (propulsion stopped), the engine continuously learns vessel-specific leeway and speed-loss corrections against GPS ground truth. There is nothing to activate: the "shadow boat" runs at all times, and the uncertainty polygon tightens over a season as the calibration bins fill up. Early on, expect conservative uncertainty estimates.
4. Optional integrations, configured in the plugin settings:
   • signalk-logbook — write confirmed fixes and maneuvers to the vessel's logbook
   • signalk-polar-performance-plugin — polar-derived speed when the paddlewheel is fouled or silent
   • @meri-imperiumi/signalk-autostate — keeps the navigation state (anchored, sailing, motoring) accurate, which gates training
   • Signal K Weather API — point-forecast current vectors for the DR solution

Daily use

Open the webapp from your Signal K server's Web apps menu. It shows the vessel and the shadow boat on the chart, GPS↔DR divergence, the uncertainty polygon, and the water-track log. A status tile is also available for Freeboard-SK ≥ 3.0.

Entering fixes. Open ⊕ Sight / LOP and pick the method:

• Bearing — compass bearing to a known object (lighthouse, tower) gives a line of position
• Vert. Angle — vertical angle to a known object of known height gives a circular position line
• Celestial — sextant sight of the Sun, Moon, or an almanac star

You can right-click an object on the map to pre-fill its coordinates. Sights can be entered with a stopwatch delay ("minutes ago") instead of clock time. Observations accumulate in the pending list; select two or more, Preview selected to resolve the candidate fix, then Confirm fix to snap dead reckoning to it. A single line of position can also be advanced to a later sight (running fix).

When GPS looks wrong. The plugin raises notifications for GPS anomalies, growing DR divergence, and paddlewheel fouling — but it never switches navigational authority by itself. If you decide to trust DR over GPS, press Engage OVERRIDE; the shadow boat takes over navigation.position instantly (it has been running all along). Release it when position is re-established, e.g. after confirming a fix.

Other controls. ≋ Current sets a manual set-and-drift override with a TTL, which outranks the automatic current sources while it lasts.