Computer-implemented method and system for planning the behavior of a participant in a traffic scene

A computer-implemented system for behavior planning of at least one participant in a traffic scene, comprising:

• a perceptual module configured to generate a grid-based scene representation based on aggregated scene-specific information;
• a pre-processing component configured to split the grid-based scene representation into multiple tiles, each tile representing a partial area of the traffic scene;
• a deep learning architecture configured to map the grid-based scene representation to latent features using the tiles and taking into account semantic relationships between individual tiles of the grid-based scene representation;
• and
• at least one downstream deep learning module configured to predict a development of the traffic scene and/or to behavior plan based on the latent features,
• wherein the perceptual module is configured to distribute the scene-specific information to at least two semantic levels of the grid-based scene representation,
• wherein the pre-processing component is configured to maintain the distribution of the scene-specific information to the at least two semantic levels during the division into the tiles and to form at least one type of sub-tile,
• wherein the sub-tiles of a type each comprise a type-defining subset of the at least two semantic levels of a tile, and
• wherein the deep learning architecture is configured to take into account semantic relationships between the sub-tiles of at least one type when mapping the grid-based scene representation to the latent features.