Detection of driving-relevant situations at a greater distance

摘要

Method (100) for recognizing at least one relevant region (12) in the environment (11) of an ego vehicle (1), in which a situation (13) relevant to the travel and / or safety of the ego vehicle (1) is present, from measurement data (21a) of at least one sensor (21) that observes at least part of the surroundings (11), the measurement data (21a) being discretized in pixels or voxels and / or otherwise suitably represented, the presence of the relevant situation (13 ) depends on the presence of at least one characteristic object (31) in the environment (11) and the resolution of the pixels, voxels or the other representation is not sufficient for direct recognition of the characteristic object (31), the measurement data (21a) being based on the The presence of a grouping (3) of objects (31-33), in which the characteristic object (31) is contained, is examined (110), the resolution of the pixels, voxels or the other representation n is sufficient for the recognition of the grouping (3) and a region in which the grouping (3) is recognized is evaluated as a relevant region (12) (120), associated classifier module (5) and data set with learning input data and learning -Output data for the classifier module.