Task 9 — Telegram Bot

Overview

The Telegram bot sends real-time alerts when threats are detected by the YOLO camera on Pi 4. Users can also query system status via commands.

Tech Stack & Packages

Package	Purpose	Install
python-telegram-bot	Telegram Bot API	pip install python-telegram-bot==20.7
requests	Call backend REST API	pip install requests

Setup Steps

Step 1 — Create a Bot

1. Open Telegram → search @BotFather
2. Send /newbot
3. Name it: EdgeComputingMonitor
4. Username: edge_monitor_bot
5. Save the TOKEN BotFather gives you

Step 2 — Get Your Chat ID

1. Start a chat with your bot
2. Send any message to it
3. Open: https://api.telegram.org/bot<TOKEN>/getUpdates
4. Find "chat":{"id": XXXXXXX} — that is your Chat ID

Bot Commands

Command	What it does
/start	Welcome message + list of commands
/status	Show all nodes with CPU, RAM, temperature
/latest	Show latest detection with image
/alerts	Show recent threat alerts
/stats	Show today's detection statistics
/help	Show all available commands
/myid	Show the caller's Telegram chat ID

Data the Bot Gets from the Backend

/status — calls GET /api/cluster/nodes

{
  "nodes": [
    {
      "name": "raspberrypi",
      "role": "control-plane",
      "status": "Ready",
      "cpu_percent": 45.2,
      "memory_percent": 62.1,
      "temperature": 52.3,
      "uptime": "10 hours, 30 minutes"
    },
    {
      "name": "pi3-1",
      "role": "worker",
      "status": "Ready",
      "cpu_percent": 12.5,
      "memory_percent": 35.0,
      "temperature": 48.1,
      "uptime": "10 hours, 30 minutes"
    }
  ]
}

/latest — calls GET /api/detections/latest

{
  "id": 42,
  "timestamp": "2026-06-16T14:30:00.000Z",
  "type": "person",
  "threat": false,
  "confidence": 0.95,
  "count": 3,
  "location": "Zone A",
  "classes": ["person", "smoke"],
  "image_url": "http://192.168.50.1:9000/detections/detection_42.jpg"
}

/alerts — calls GET /api/alerts

{
  "alerts": [
    {
      "id": 41,
      "timestamp": "2026-06-16T14:29:58.000Z",
      "severity": "high",
      "message": "Threat at Zone B (knife)",
      "acknowledged": false
    }
  ]
}

/stats — calls GET /api/stats

{
  "total_detections_today": 142,
  "total_people_detected": 139,
  "total_threats_detected": 3,
  "detection_accuracy": 94.5,
  "active_nodes": 6,
  "total_nodes": 7
}

What Messages the Bot Sends

Automatic Threat Alert (sent every 30 seconds if a new threat is found)

⚠️ THREAT DETECTED!

📍 Location: Zone B
🔍 Object: knife
📊 Confidence: 87%
🕐 Time: 14:29:58
🆔 Detection ID: #41

[Send the detection image here]

/status response

🖥️ Cluster Status

✅ raspberrypi (Master)
   CPU: 45.2% | RAM: 62.1% | Temp: 52.3°C

✅ pi3-1 (Worker)
   CPU: 12.5% | RAM: 35.0% | Temp: 48.1°C

✅ pi3-4 (Worker)
   CPU: 8.3% | RAM: 28.5% | Temp: 45.6°C

Active: 6/7 nodes

/latest response

🎯 Latest Detection

Type: Person
Confidence: 95%
Count: 3
Location: Zone A
Classes: person, smoke
Time: 14:30:00

[Send the detection image here]

/stats response

📊 Today's Statistics

👁️ Total Detections: 142
👤 People Detected: 139
⚠️ Threats Detected: 3
🎯 Accuracy: 94.5%
🖥️ Active Nodes: 6/7

/alerts response

🔔 Recent Alerts

🔴 Threat at Zone B (knife)
   Time: 14:29:58 | NOT acknowledged

🔴 Threat at Zone A (gun)
   Time: 14:25:10 | ✓ Acknowledged

Important Logic

Threat Detection Loop

• Run a background task every 30 seconds
• Call GET /api/alerts
• For each alert where acknowledged = false → send a Telegram message
• Keep track of which alert IDs have already been sent (avoid duplicates)

Sending Images

• Get image_url from detection data
• Download the image from the URL
• Send it as a photo to the Telegram chat

Backend URL

http://192.168.50.1:5000

Must be connected to the cluster network (192.168.50.x).

File Structure

telegram-bot/
├── bot.py          ← Main bot file
├── config.py       ← BOT_TOKEN, CHAT_ID, BACKEND_URL
└── requirements.txt

Development Tips

1. Test commands one by one — start with /start
2. Use mock data first — don't need the cluster to start
3. Add the threat alert loop last
4. Keep BOT_TOKEN secret — never commit to GitHub
5. Run on Pi 5 for 24/7 operation:

python3 bot.py

---

Frankfurt University of Applied Sciences — Cloud Computing SS2026 — Group 8