A computing system predicts where a gaze of a vehicle operator (“driver”) will be in the near future to infer driver attention rather than inferring driver attention based on currently detected gaze data, which may be stale by the time the computing system processes the data and infers attention. The gaze may be predicted based on data from a camera that captures an image of the driver’s eyes, head, posture, or other aspects. In some examples, the computing system uses a machine-learned model (ML model), such as a recurrent neural network (RNN) model to predict future glance behavior. The machine-learned model is trained using machine learning. The computing system may provide the predicted future driver gaze or glance information to application modules requiring driver gaze information, such as applications that monitor driver gaze and infer driver attention. In some examples, the computing system may use the gaze information based on the gaze prediction to take an action, such as to prompt the driver to return attention to the driving task.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Keller, Michael, "PREDICTING FUTURE GLANCE BEHAVIOR FROM GAZE PATTERN HISTORY", Technical Disclosure Commons, (July 15, 2019)