Computer-implemented method and system for training an AI-based prediction model

Abstract

Proposed is a computer-implemented method and system for training an AI-based prediction model (10) for a given behavior planner (11), which plans the future behavior of an at least partially automated ego vehicle based on aggregated scene-specific information. The prediction model (10) is trained to predict the future development of a traffic scene based on aggregated scene-specific information. At least one training dataset with training data elements i, generated from scene-specific information of training scenes, is used for training. According to the invention, a weight is determined for each training data element i using the behavior planner (11), which determines the extent to which the respective training data element i is considered during the training of the prediction model (10).