Send Depth node

send_depth is an edge-only transform node that reads a float depth map from an upstream call_model node (any registered depth estimator — Hailo fast_depth / SCDepthV3, Video-Depth-Anything) and publishes it as a uint16 H×W payload on the canonical twin depth topic cyberwave/twin/{twin_uuid}/depth — the same shape the PointCloud viewer, recording / replay and the object_pose node already consume.

Typical chain

trigger (camera_frame)
  → call_model  (depth-estimation model)
  → send_depth

Pipe the same call_model output into object_pose when the model is metric to lift 2D detections into 3D positions.

Inputs

Field	Type	Required	Description
`depth_map`	array	✓	H×W float32 depth map. Auto-wired when placed downstream of a depth-estimation `call_model`.
`twin_uuid`	string	✓	Target twin — depth is published on `cyberwave/twin/{twin_uuid}/depth`.

Parameters

Field	Default	Description
`output_mode`	`normalized_uint16`	`normalized_uint16` stretches the depth range to 0-65535 for visualisation. `metric_mm` multiplies by `scale_factor` and clips — required by consumers that need real distances.
`scale_factor`	`1000.0`	Applied only in `metric_mm` mode. Metres → millimetres by default.
`publish_every_n_frames`	`1`	Throttle. Set higher on tight uplinks.

Notes

Edge-only. The node is filtered out of cloud workflows at compile time.
Replaces the retired cyberwave-edge-camera-depth-estimation driver — the same MQTT topic shape, now expressed as a workflow.

Object Pose node Spatial Filter node

Concepts

Platform Features

Cyberwave Edge

Technical Reference

Use-Case Recipes

Typical chain

Inputs

Parameters

Notes

​Typical chain

​Inputs

​Parameters

​Notes

Typical chain

Inputs

Parameters

Notes