Real-Time Fleet Intelligence: How ASI Biont Integrates with GPS/GLONASS Trackers via NMEA and COM Port

The Problem: Blind Spots in Fleet Operations

For logistics companies and field service teams, GPS trackers are the backbone of operational visibility. However, most commercial tracking platforms are closed ecosystems: they offer a dashboard but little to no programmatic control. When a dispatcher needs to correlate a sudden temperature spike in a refrigerated truck with a specific GPS coordinate, or trigger an alert when a vehicle enters a geofenced zone after hours, they hit a wall. Standard interfaces require manual polling, custom middleware, or expensive API subscriptions.

Traditional NMEA 0183 GPS receivers output sentences like $GPGGA (fix data) and $GPRMC (recommended minimum) over a serial COM port at 4800 or 9600 baud. The data is there—latitude, longitude, speed, heading, time—but extracting and acting on it in real time typically demands a dedicated server, a polling script, and a developer to glue it all together.

The Solution: ASI Biont + Serial Bridge

ASI Biont connects to any NMEA-capable GPS/GLONASS tracker via a Hardware Bridge running on a local PC (Windows, Linux, macOS). The bridge opens a COM port (e.g., COM3 or /dev/ttyUSB0) at the tracker’s baud rate and establishes a WebSocket connection to the ASI Biont cloud. The AI agent then uses the industrial_command tool with the serial_write_and_read(data=hex_string) method to send commands and parse responses.

Why this method? Because NMEA is a text-based protocol over RS-232. The bridge handles low-level serial I/O (including Windows overlapped I/O recovery via CancelIoEx and PurgeComm), while the AI handles parsing, decision-making, and integration with other services (email, Slack, databases).

Concrete Use Case: Real-Time Route Monitoring for a Cold-Chain Fleet

A mid-sized dairy distributor in the Midwest runs 40 refrigerated trucks. Each truck has a Teltonika FMB920 GPS tracker (GLONASS + GNSS) configured to output NMEA sentences over RS-232 at 9600 baud. The dispatcher needs to:
- Track each truck’s position every 30 seconds.
- Detect when a truck leaves its assigned delivery zone.
- Log coordinates to a PostgreSQL database for later analysis.
- Send an SMS alert if a truck stops for more than 10 minutes outside a depot (possible theft or breakdown).

Step-by-Step Integration with ASI Biont

1. Setup the Hardware Bridge

The user downloads bridge.py from the ASI Biont dashboard (Devices → Create API Key → Download bridge). They run it on a Windows machine at the dispatch office:

pip install pyserial requests websockets
python bridge.py --token=YOUR_API_TOKEN --ports=COM3 --baud 9600 --rate=10

The bridge opens COM3 at 9600 baud and connects to ASI Biont via WebSocket. The --rate=10 flag limits command execution to 10 per second to avoid overwhelming the tracker.

2. AI Connects to the Tracker

In the ASI Biont chat, the user types:

“Connect to the GPS tracker on COM3 at 9600 baud. Continuously read NMEA sentences, parse $GPGGA and $GPRMC, and log latitude, longitude, speed, and UTC time to a PostgreSQL table named gps_log. Also, send me a Telegram alert if any truck’s speed exceeds 80 km/h.”

3. What the AI Does Behind the Scenes

The AI generates and executes a Python script in the sandbox (via execute_python). The script uses pynmea2 to parse NMEA sentences and psycopg2 to log data. It does not run while True (sandbox timeout is 30 seconds). Instead, it polls the bridge once, processes a batch of sentences, and exits. For continuous monitoring, the user sets up a cron job (or Windows Task Scheduler) that calls the AI endpoint every 30 seconds.

Here is a simplified example of the script the AI might write:

import pynmea2
import psycopg2
import requests

# Simulate reading raw NMEA from bridge via industrial_command
# In reality, the AI sends industrial_command(protocol='serial', command='serial_write_and_read', data='...')
raw_nmea = "$GPGGA,123519,4807.038,N,01131.000,E,1,08,0.9,545.4,M,46.9,M,,*47\n$GPRMC,123519,A,4807.038,N,01131.000,E,022.4,084.4,230394,003.1,W*6A"

for line in raw_nmea.split('\n'):
    if line.startswith('$GPGGA') or line.startswith('$GPRMC'):
        msg = pynmea2.parse(line)
        if isinstance(msg, pynmea2.GGA):
            lat = msg.latitude
            lon = msg.longitude
            alt = msg.altitude
            utc = msg.timestamp
        elif isinstance(msg, pynmea2.RMC):
            speed = msg.spd_over_grnd  # knots
            speed_kmh = speed * 1.852
            if speed_kmh > 80:
                requests.post("https://api.telegram.org/bot<TOKEN>/sendMessage",
                              json={"chat_id": "<CHAT_ID>", "text": f"Speed alert: {speed_kmh:.1f} km/h"})
        # Log to PostgreSQL
        conn = psycopg2.connect("dbname=fleet user=admin password=secret host=192.168.1.50")
        cur = conn.cursor()
        cur.execute("INSERT INTO gps_log (lat, lon, speed, utc) VALUES (%s, %s, %s, %s)",
                    (lat, lon, speed_kmh, utc))
        conn.commit()
        cur.close()
        conn.close()

4. Results Achieved

5. Why This Works

The bridge’s serial_write_and_read atomic operation ensures that every read is preceded by a write, preventing buffer corruption common with multi-threaded serial access. On Windows, if a write fails (returns 0 bytes), the bridge automatically runs CancelIoEx to cancel pending overlapped operations, then PurgeComm to clear RX/TX buffers, then retries with synchronous WriteFile. This makes the connection robust even with flaky USB-to-serial adapters.

Beyond GPS: Any Serial Device, Any Protocol

The same pattern works for any device that speaks over a COM port: Arduino, ESP32, Modbus RTU, barcode scanners, weigh scales, and even legacy CNC machines. The user only needs to describe the device and the expected data format in the chat. ASI Biont handles the rest.

How to Get Started

No coding required. Go to asibiont.com, create an API key, download the bridge, and start a chat. Tell the AI:

“I have a GPS tracker on COM4 at 9600 baud. Parse NMEA and log coordinates to a Google Sheet.”

The AI will generate the integration script, connect to your device, and begin monitoring in seconds.

Conclusion

Integrating GPS/GLONASS trackers with an AI agent like ASI Biont turns raw NMEA sentences into actionable intelligence—without middleware, without a dedicated backend, and without waiting for a developer. The combination of a lightweight serial bridge and a generative AI that writes and executes integration code on the fly makes fleet monitoring, geofencing, and alerting accessible to any organization, regardless of technical depth.

Try it today at asibiont.com.