🛴 Realtime Xiaomi Scooter Monitoring and Control using a Raspberry Pi, GPS and 4G
An open-source IoT solution designed to monitor and control your Xiaomi e-scooter in real-time using a Raspberry Pi, GPS, and 4G connectivity.
This project is my Computer Science final year project at the University of A Coruña.
Disclaimer: Although the project is fully functional, the code quality may not be top-tier, as this was my first time coding in Rust. My background is in Systems Administration and Infrastructure, not software development.
The project has been tested on the Xiaomi Mi Pro 2 scooter but should be compatible with the following models: m365, Mi Lite 1S, Mi Pro, Mi Pro 2, and Mi Pro 3.
Part of the code comes from the m365 repository and the CamiAlfa Reverse Engineering work.
-
Real-Time Data Monitoring using Grafana: View live telemetry from your e-scooter, including speed, trip distance, uptime, total odometer, estimated range left, battery status (capacity, percentage, voltage, current, temperature), and GPS location via a Grafana dashboard.
-
Remote Control: Disable your scooter remotely using a relay.
-
LTE and GPS Connectivity: Stay connected wherever you go, with 4G modem and GPS integration.
-
Third Party integration: Integrate your scooter with platforms like Home Assistant using the API.
-
Own your data: All infrastructure is self-hosted, allowing you to maintain complete ownership of your data.
-
Automatically switch between 4G and Wi-Fi: Save up on your data plan by using Wi-Fi when it is available.
🛠️ Hardware:
- Raspberry Pi Zero 2W
- Waveshare SIM7600G-H 4G HAT
- 5V Relay
- Xiaomi Dashboard Cable Connector
- SIM Card for LTE connection
- SD Card for Raspberry Pi
- External Battery
- Electrical register box.
I attempted to power the Raspberry Pi using the scooter’s battery, but due to its high power consumption, this was not feasible. It may be possible with a microcontroller.
Why did I use a Raspberry Pi?: Given my tight deadline (2.5 months over the summer), I chose the Raspberry Pi because it supports Rust, allowing me to reuse parts of the m365 library. Although a microcontroller would be better suited due to the lower energy consumption, it would have required rewriting the code in C, which was not feasible in such a short time frame.
💻 Software:
- Raspberry Pi OS Lite 64-bit
- Python (Paho-MQTT, FastAPI, SQLAlchemy)
- Grafana (Data visualization)
- TimescaleDB (PostgreSQL extension for time series data) + PostGIS
- Docker and Ansible (Server deployment)
- Tailscale (Optional VPN for secure access)
-
Deploy the server. Follow the server documentation.
-
Compile the client (optional) and obtain the scooter's MAC and pairing token: Detailed instructions can be found in the client guide
-
Set up the Raspberry Pi client and start it Connect the relay to the scooter and install the monitoring client. For further details, refer to the Raspberry setup guide.
I enjoy tinkering with Raspberry Pi (and other hardware) and data visualization. A few years ago, I used a Bird rental scooter and was curious about how it worked. When I needed to choose a topic for my thesis, I remembered that experience and decided to build my own scooter monitoring system.
While researching, I found very few DIY projects that were user-friendly, so I aimed to create something accessible for anyone interested in monitoring and using their scooter's data.
-
Secure the API: Due to time constraints, the API currently lacks security measures. Future updates might include instructions on setting up an API Gateway or other security methods.
-
Add a reverse proxy to expose the server components securely on the internet without a VPN.
-
Add support for multiple users and scooters on the same server.
-
Add support for other types of vehicles (cars, bikes, other scooter models).
-
Obtain data using the scooter's serial connection using a cable to avoid possible Bluetooth interference.
-
Reimplement the client on a microcontroller for lower power consumption.
If you have suggestions, bug fixes, or ideas for improvements, feel free to open an issue or submit a pull request. Contributions are welcome!
This project uses parts of the code from the m365 library to handle cryptography, connection and data collection from the scooter. These portions of the code remain mostly unchanged, with minor fixes for data conversion, as detailed in this Github Pull Request.
The project also relies on the reverse-engineered protocol from CamiAlfa.
I am deeply grateful to both developers for their work, which was essential to the success of this project.
This project is licensed under the terms of the AGPL-3 license.