Skip to main content

Cyberwave is in Private Beta.

Request early access to get access to the Cyberwave dashboard.

SDK calls for managing edge devices, publishing GPS telemetry, and reading the time-aware data bus. The deep concepts behind edge workers and data fusion live on the Edge & Drivers pages — this page covers the SDK surface and links out.

​
Edge management

Manage edge devices (Raspberry Pi, Jetson, etc.) that run the Cyberwave Edge Core: 
cw = Cyberwave()

edges = cw.edges.list()
for edge in edges:
    print(edge.uuid, edge.name, edge.fingerprint)

edge = cw.edges.get("your-edge-uuid")

edge = cw.edges.create(
    fingerprint="linux-a1b2c3d4e5f60000",
    name="lab-rpi-001",
    workspace_id="your-workspace-uuid",
    metadata={"location": "lab-shelf-2"},
)

edge = cw.edges.update(edge.uuid, {"name": "lab-rpi-001-renamed"})
cw.edges.delete(edge.uuid)
The fingerprint is a stable hardware identifier derived from hostname, OS, architecture, and MAC address. The Edge Core generates and persists it automatically at ~/.cyberwave/fingerprint.json on first boot.

Edge Workers

The edge worker runtime, hook system, and how drivers run on-device.

​
GPS telemetry

Publish GPS fixes for a twin. Outbound GPS is rate-limited to 2 Hz per twin; a fix_type="none" update is dropped without consuming the rate slot. 
twin = cw.twin(twin_id="your-twin-uuid")
twin.update_twin_gps(latitude=37.7749, longitude=-122.4194, fix_type="rtk")
See Outbound MQTT rate limits for the full throttling contract.

​
Data layer (time-aware)

The cw.data data bus carries time-stamped samples (frames, joints, GPS, custom channels) between drivers and consumers. Time-aware fusion and synchronized hooks let you align samples across channels by timestamp rather than arrival order.
Experimental: the basic cw.data pub/sub surface is under active development and may change. The fusion and synchronized-hook primitives below are documented on the Edge & Drivers pages — use those as the source of truth.

Data Fusion

Time-aware fusion primitives: point reads, interpolation, and time-window queries.

Synchronized Hooks

Run a callback when multiple channels have aligned samples.

​
Automatic detection publishing

Edge ML workers can publish detections back onto the data bus so downstream nodes and the platform can consume them. The worker runtime and detection routing are covered under Edge Workers and ML Models.