Processing of radar signals including suppression of motion artifacts

A method for evaluating moving objects from radar signals, comprising the following steps:

• determining a processed representation of the scene observed using radar signals, using a trainable module trained by:
• feeding actual radar signals of a scene and/or actual representations derived from the actual radar signals of the scene observed, using the actual radar signals, to the trainable module, and converting, by the trainable module, the actual radar signals and/or the actual representation to processed radar signals of the scene and/or to processed representations of the scene,
• using a cost function to assess to what extent the processed radar signals are suited for reconstructing a movement of objects or to what extent the processed representations contain artifacts of moving objects in the scene, and
• optimizing parameters, which characterize performance characteristics of the trainable module, to receive processed radar signals or processed representations, which are better assessed by the cost function;
• determining a second representation of the scene observed using the radar signals from input fed to the trained module;
• and
• determining spatially resolved information about a velocity of objects included in the representations from a difference between the processed representation and the second representation.

A method, comprising the following steps:

• training a trainable module by:
• feeding actual radar signals of a scene and/or actual representations derived from the actual radar signals of the scene observed, using the actual radar signals, to the trainable module, and converting, by the trainable module, the actual radar signals and/or the actual representation to processed radar signals of the scene and/or to processed representations of the scene,
• using a cost function to assess to what extent the processed radar signals are suited for reconstructing a movement of objects or to what extent the processed representations contain artifacts of moving objects in the scene, and
• optimizing parameters, which characterize performance characteristics of the trainable module, to receive processed radar signals or processed representations, which are better assessed by the cost function;
• measuring radar signals from a driving environment of a vehicle using at least one radar sensor mounted on a vehicle;
• evaluating the radar signals using the trained module;
• generating at least one control signal from s result of the evaluating;
• and
• controlling the vehicle using the control signal.

A non-transitory machine-readable data carrier on which is stored a computer program for training a trainable module for evaluating radar signals, the computer program, when executed by a computer, causing the computer to perform the following steps:

• feeding actual radar signals of a scene and/or actual representations derived from the actual radar signals of the scene observed, using the actual radar signals, to the trainable module, and converting, by the trainable module, the actual radar signals and/or the actual representation to processed radar signals of the scene and/or to processed representations of the scene;
• using a cost function to assess to what extent the processed radar signals are suited for reconstructing a movement of objects or to what extent the processed representations contain artifacts of moving objects in the scene;
• and
• optimizing parameters, which characterize performance characteristics of the trainable module, to receive processed radar signals or processed representations, which are better assessed by the cost function.

A computer configured to train a trainable module for evaluating radar signals, the computer configured to:

• feed actual radar signals of a scene and/or actual representations derived from the actual radar signals of the scene observed, using the actual radar signals, to the trainable module, the trainable module converting the actual radar signals and/or the actual representation to processed radar signals of the scene and/or to processed representations of the scene;
• use a cost function to assess to what extent the processed radar signals are suited for reconstructing a movement of objects or to what extent the processed representations contain artifacts of moving objects in the scene;
• and
• optimize parameters, which characterize performance characteristics of the trainable module, to receive processed radar signals or processed representations, which are better assessed by the cost function.