The Configuration File

FusionHub runs the pipeline described in a single JSON configuration file, loaded with -c path/to/config.json. The node editor and this file are two views of the same thing: what you wire in the editor is what is written to the file, and a hand-edited file appears in the editor the same way. The usual workflow is to build the graph in the editor and hand-edit only when scripting or deploying headless.

File layout

{
  "settings": { "explicitConnections": true },
  "sources": { },
  "sinks": { }
}

• settings - global settings for the whole pipeline.
• sources - all source nodes.
• sinks - all filter and sink nodes. Both live in this section; the node’s registered role decides how it is wired.

Node entries

Each entry in sources / sinks is keyed by an instance name of your choice.

• If the key is a known node type ID (for example gnssImuFusion), it selects the node type directly. Any name works too; then set the type with "nodeTypeId": "gnssImuFusion". The config aliases listed on each node’s documentation page are accepted as well.
• "type" selects a subtype for nodes that have them (see the Subtypes section on the node’s page).
• "enabled": false disables a node without removing it from the file.
• "settings" holds only the values you want to change. Every setting is optional: anything omitted uses the default listed in the Properties table on the node’s documentation page. The keys are exactly the code keys shown there.

Wiring

Data flows over named endpoints:

• A producing node binds one endpoint: "outEndpoint" on sources, "dataEndpoint" on filters. Omit it and the runtime generates one.
• A consuming node lists the endpoints it reads from in "inputEndpoints".
• "inputDataFilter" optionally restricts which data types the node accepts from those endpoints; anything not in the list is dropped.
• With "explicitConnections": true in the global settings (the editor always writes this), only the listed connections exist. When it is absent or false, a node without "inputEndpoints" subscribes to every source - a legacy convenience for tiny hand-written configs.

The editor may additionally write "inputConnectionTypes", "inputSlotAssignment" and "inputRoutes" to pin per-connection data types and input slots, plus a "_layout" block for node positions. Treat these as editor-managed and leave them alone when hand-editing.

Example

A small pipeline: GNSS receiver and IMU fused by GNSS-IMU Fusion, with the result written to a log file.

{
  "settings": { "explicitConnections": true },
  "sources": {
    "gnss": {
      "outEndpoint": "inproc://gnss_data",
      "settings": { "port": "COM5" }
    },
    "lpms": {
      "outEndpoint": "inproc://imu_data",
      "settings": { "name": "ig1232800650" }
    }
  },
  "sinks": {
    "gnssImuFusion": {
      "dataEndpoint": "inproc://fused_data",
      "inputEndpoints": ["inproc://gnss_data", "inproc://imu_data"],
      "settings": { }
    },
    "logger": {
      "inputEndpoints": ["inproc://fused_data"],
      "settings": { "filePath": "drive.json" }
    }
  }
}