A computer-implemented method trains a machine-learning model to assess accident warning levels for two- or three-wheeled vehicle drivers using automatically acquired sensor data.

Patentradars sammanfattning

Autoliv Development AB discloses a training methodology for a machine-learning model targeting accident warning level assessment specifically for two-wheeled and three-wheeled vehicles (e.g., motorcycles, trikes). The method systematically prepares training data by automatically acquiring multiple sets of sensor data collected over corresponding time periods, then associates each dataset with a discrete warning level drawn from a predefined plurality of levels. The pipeline further maps sensor data to hazard levels, links those hazard levels to configurable hazard thresholds, and allows setting or adjusting those thresholds, enabling nuanced, data-driven risk classification tailored to the unique dynamics of narrow-track vehicles.

AI-genererad och redaktionellt processad.