COMPREDICT’s Virtual Headlight Leveling Sensor on Sonatus AI Director — Unlocking In-Vehicle AI for Smarter Headlight Leveling

Traditionally, automatic headlight leveling requires one or two physical ride height sensors installed on the vehicle chassis to estimate pitch angle. COMPREDICT replaces this hardware with a software-only AI model that uses existing CAN bus signals—such as acceleration, wheel speeds, and suspension behavior—to accurately estimate the vehicle’s pitch. This virtual approach eliminates the need for costly physical components installation, calibration, and maintenance.

To meet both regulatory and real-world driving needs, the AI model generates two types of pitch angle estimations:
- Static Pitch Angle: Used for regulatory certification scenarios, such as testing the vehicle under loaded versus unloaded conditions.
- Dynamic Pitch Angle: Provides real-time responsiveness during active driving situations like acceleration, braking, driving on slopes, or traversing bumpy roads.

Integrating new software across different vehicle platforms can take months. Sonatus AI Director reduces this timeline to days by providing a standardized, robust runtime environment. It streamlines the process by:
- Standardizing Data Access: Offering unified APIs to easily access necessary vehicle signals (like CAN data) across different architectures.
- Protecting IP: Running the model in a secure, containerized environment to safeguard COMPREDICT's proprietary algorithms.
- Optimizing Performance: Automatically matching the model’s performance to the available compute resources of the specific target ECU.