Controlling Hardware: MQTT-Driven LED Automation on AWS Greengrass v2 — SciTech Edge Advance 1.1

Executive Summary¶

In our journey so far, we've established local IPC and a cloud MQTT bridge. But an Edge Device without I/O is just a server. Today, we break the software barrier and control physical hardware.

This log documents the integration of the LGPIO library on the Raspberry Pi CM5 to control LEDs via MQTT commands from the cloud. We address the critical "Permission Denied" errors associated with hardware access from inside a Greengrass container.

Hardware Stack¶

• Edge Device: Raspberry Pi CM5
• GPIO 17 (Pin 11): Red LED (Output)
• GPIO 15 (Pin 10): Green LED (Output)
• Circuit: Standard LED setup with current-limiting resistors.

Software & Tooling Stack¶

• AWS IoT Greengrass v2 Core
• LGPIO Library (Linux GPIO for ARM)
• System Python Packages (No virtualenv isolation)

Technical Walkthrough¶

The Logic: Cloud -> MQTT -> GPIO

The workflow is straightforward but powerful:

1. Cloud: User publishes a JSON command (e.g., {"red": 1, "green": 0}) to cmd/led.
2. Core: Greengrass receives the message via the MQTT Bridge.
3. Component: Python script parses the payload and toggles the GPIO pins using lgpio.

The Code (led_control.py)

We use lgpio instead of the older RPi.GPIO because the CM5 and newer kernels require standard Linux GPIO character device access.

import lgpio
import json
import awsiot.greengrasscoreipc

# Chip 0 is usually the default for Pi 4/CM4/CM5

h = lgpio.gpiochip_open(0)

# Setup Pins

RED_PIN = 17
GREEN_PIN = 15
lgpio.gpio_claim_output(h, RED_PIN)
lgpio.gpio_claim_output(h, GREEN_PIN)

def on_message_received(event):
payload = json.loads(event.message.decode('utf-8'))

  if "red" in payload:
      lgpio.gpio_write(h, RED_PIN, payload["red"])

  if "green" in payload:
      lgpio.gpio_write(h, GREEN_PIN, payload["green"])

The "Rogue Insight"

Recipe Configuration (v1.0.9)

We updated the recipe to version 1.0.9. Note that we are not installing lgpio in the Install lifecycle script anymore. We assume the environment is prepared.

Verification¶

We validated the control using the AWS IoT Core Test Client.

Topic: cmd/led

Payload 1 (Red ON):

{
  "red": 1
}

Result: Red LED lights up instantly.

Payload 2 (Both ON):

{
  "red": 1,
  "green": 1
}

Result: Both LEDs active.

Payload 3 (All OFF):

{
  "red": 0,
  "green": 0
}

Result: Complete blackout.

Closing Note¶

Connecting a lightbulb to the internet is the "Hello World" of IoT, but doing it reliably via a managed edge runtime like Greengrass is a significant engineering milestone. We now have a secure, managed pipeline to control physical infrastructure from the cloud.

— DK Swami Founder, Scitech Industries