Deploy edge container nodes, subscribe to MQTT defect streams, and fetch real-time analytics JSON payloads.
Run the SightQC model classifier container locally at the edge. Mount your hardware GPU acceleration layer (such as NVIDIA CUDA runtime) and direct local digital camera video capture units (`/dev/video0`).
docker run -d --name sightqc-edge-node \ --restart=always \ --device=/dev/video0:/dev/video0 \ --gpus all \ -e AWS_REGION=ap-south-1 \ -e EDGE_LINE_ID=mumbai_line_3 \ -e IOT_THING_NAME=mumbai_actuator_node \ registry.arcaisystech.com/sightqc/edge-runtime:v1.3.2
Local containers publish JSON payloads to your edge broker. Subscribe to the topic /sightqc/line_3/defects to trigger rejection actuators.
{
"timestamp": "2026-08-05T21:26:00Z",
"line_id": "mumbai_line_3",
"product_segment": "SMT_PCB_247",
"classification": {
"status": "DEFECTIVE",
"class": "missing_resistor",
"confidence": 0.987,
"coordinates": [142, 85, 210, 120]
},
"actuator_trigger": true
}
Subscribe to edge triggers and actuate digital GPIO pins in python. Write customized loops for local conveyor PLC boards.
import paho.mqtt.client as mqtt
import RPi.GPIO as GPIO
GPIO.setmode(GPIO.BCM)
GPIO.setup(18, GPIO.OUT) # Pneumatic actuator pin
def on_message(client, userdata, msg):
data = json.loads(msg.payload)
if data["classification"]["status"] == "DEFECTIVE":
GPIO.output(18, GPIO.HIGH) # Trigger rejection
time.sleep(0.1)
GPIO.output(18, GPIO.LOW)
client = mqtt.Client()
client.connect("localhost", 1883)
client.subscribe("/sightqc/+/defects")
Export audit logs programmatically. Query analytics endpoints to extract defect rates for supplier review.
# Fetch hourly defect rate statistics curl -X GET "https://api.arcaisystech.com/v1/mumbai_line_3/analytics?range=24h" \ -H "Authorization: Bearer $ARCAISYS_API_TOKEN"