Skip to content

Public-safe docs: Examples use placeholders. Live fleet IDs, credentials, telemetry snapshots, harness evidence, and session chronology are maintained in operator-controlled handoff repositories — not on the public web.

About this article

  • Audience: Developers and field engineers setting up a Sentinel Field Kit for the first time
  • Goal: Boot the local Shadow camera demo and verify the real-time inference stream in under 3 minutes
  • Type: Quickstart

Summary

This quickstart guides you through the physical connection and software initialization of the Sentinel Field Kit. Follow these three steps to deploy the local camera environment and confirm pipeline readiness.

Prerequisites

  • Sentinel Field Kit (including Shadow camera and Host Edge Device)
  • High-speed USB-C to USB-A/C cable (USB 3.0 compatible)
  • SSH client on your development workstation
  • Local network router with DHCP enabled

For the full COTA POC desk (Mac Studio BOM, UPS, Ollama, stack ports), use Host materials — Mac Studio instead of this minimal field insert.

1. Connect the hardware

Mount the Shadow camera to its tripod or bracket, then connect it to the Host Edge Device.

  • USB 3.0 connection: Connect the USB-C cable to the Shadow camera, and plug the other end into a blue USB 3.0 port on the Host Edge Device. Do not use USB 2.0 ports, as they lack the bandwidth required for high-resolution spatial AI streams.
  • System power: Power on the Host Edge Device. The status LED will pulse, indicating that the system is booting.
  • Network connection: Plug the Host Edge Device into the local router via Ethernet, or verify it has connected to the pre-configured Field Kit Wi-Fi network.

2. Boot the orchestrator stack

Establish an SSH session to the Host Edge Device and boot the local runtime environment.

  • Establish connection: Open a terminal on your development workstation and connect to the device:
bash
ssh dean@<device-ip>
# Input the pre-configured field credentials when prompted
  • Launch the stack: Navigate to the local firmware workspace and launch the demo containers:
bash
cd [operator-workspace]/oak-vms-firmware
./scripts/boot-local-demo.sh

The boot script initializes the local camera pipeline, spins up the local MQTT broker, and starts the Phantom proxy wrapper.

3. Verify stream and telemetry

Verify the system is processing video frames and publishing spatial AI detections locally.

  • Inspect proxy health: Run a curl to ensure the local Phantom proxy is responding:
bash
curl -s http://localhost:8788/health | jq .
  • Check telemetry stream: Query the local MQTT cache to verify that the camera is publishing temperature and inference data:
bash
curl -s http://localhost:8788/fleet/mqtt/latest | jq '.devices[].die_temp, .devices[].detections'
  • Launch SCION GRID HUD: Open your web browser and navigate to http://<device-ip>:9119 to view the live video canvas with overlayed tracking nodes.

Next steps

Operator depth

Full assembly schematics, camera calibration matrix logs, and board-level specifications are maintained in the private operator handoff (not published) (private).