Technology · Edge architecture
One compute node per function, one hub.
Each function gets its own node with its own sensors. The hub records every stream and shows it on the dashboard. New task? Add a node.
| Hub | Two configurations on the same board (RUBIK Pi 3, Qualcomm QCS6490) and software: Multi-Camera Hub (three global-shutter cameras, capture up to about 100 fps, recording rate set per order; radiometric thermal imager; 48 kHz audio as event recording, continuous audio built to order; NVMe storage) or Triple Eye (depth, vision, thermal) |
|---|---|
| Nodes | Vision, vibration, acoustic, depth, event, wireless |
| Time-stamping | Every stream is time-stamped on the hub. A node's clock offset is measured continuously and written into every record, with its uncertainty. Hardware sync timer in development with LarsLabs. |
| Evidence | Per-segment SHA-256 hashes, append-only journal, tiered retention; after a power loss the gap is logged, alerted and reported, and recording resumes by itself |
| Views | Operator screen straight from the edge device, on the local network; Grafana, OpenSearch and Kafka integration on request |
Why one node per function?
- Each function can be added or replaced on its own
- A fault in one node does not stop the others
- Compute grows with the task, not up front
Next step
Does the architecture fit your plant?
We work out with you which nodes your task needs and where they go.