Maximizing the Value of Vehicle Data

The traditional approach involved indiscriminately collecting massive amounts of data from all vehicle domains, which created significant waste.
- Wasted Resources: It tied up scarce ECU processing power and consumed excessive network bandwidth to upload data that was largely irrelevant .
- High Costs: The OEM incurred high costs for wireless data transmission and cloud storage. For example, finding a specific issue like braking efficiency required capturing 28TB of raw data from 1,000 cars just to isolate a few relevant incidents.
- Slow Analysis: Sifting through this massive "data haystack" to clean and organize information often took weeks or months, delaying critical improvements.

Unlike static logging, which captures everything continuously, Sonatus Collector uses event-driven policies. It allows the OEM to define precise "trigger conditions"—such as specific speeds, geofences, or fault codes—so the vehicle only collects and uploads high-quality data when those specific events occur .
- Flexibility: These policies are lightweight (measured in KBs) and can be deployed or updated instantly from the cloud without requiring complex ECU reflashing or heavy OTA software updates.
- Precision: This approach filters "signal from noise" at the edge (in the vehicle), eliminating the need to upload and process terabytes of irrelevant data.

The OEM realized an 80x overall improvement in data collection efficiency. Specific metrics include:
- Expanded Use Cases: They increased data utilization by 20x, growing from 5 to over 100 active use cases.
- Reduced Data Volume: Despite the increase in use cases, they reduced the daily data volume per vehicle by 4x, dropping from 100MB down to approximately 24MB.
- Faster Insights: For complex issues like tuning pedestrian detection algorithms, they reduced the time to identify and fix root causes from months to days.