The Challenge: A Hardware Wall
The initial goal was to extract telemetry data directly from the vehicle's OBD-II port. However, a deep analysis revealed insurmountable hardware limitations and security measures, making this path unfeasible. A completely new solution was needed.
Engineer
Hardware Access
Vehicle Data
The Architecture in Action: An Automated Flow
The solution bypasses the hardware block by creating an intelligent ecosystem that analyzes images. The process is divided into three robust and fully automated phases, turning a simple screenshot into a structured database record.
1. Ingestion
A screenshot from the "Car Scanner" app is saved to the cloud (Google Drive) and automatically synced to a local NAS, ready for processing.
2. AI Processing
A Python script sends the image to the Gemini API. The AI extracts data (SoC, SoH, Voltage), structures it into JSON, and stores it in an SQLite database, determining if the event is a trip or a charge.
3. Notification & Archiving
After saving, a Telegram bot sends a notification with key data and the screenshot for validation. The file is then archived, and cloud sync prevents any duplication.
The Results: High-Resolution Data
The final result is a continuous and reliable stream of telemetry data. This chart shows the State of Charge (SoC) over time, visualizing discharge (trips) and charge cycles, a fundamental basis for future predictive analysis.
Key Benefits
This architecture is not just a solution, but a qualitative leap in the data collection process.