Granted Patent B2

US 12,613,306 · App. 17/827,153 · Granted Apr 28, 2026

Inventors: Tai Fei (Hamm, DE); Christopher Grimm (Lippstadt, DE); Frank Gruenhaupt (Marsberg, DE); Ernst Warsitz (Paderborn, DE)

Assignee: Hella GmbH & Co. KGaA

G01S7/02G01S13/931G06N3/08

Abstract

Claims (37)

1 . A method for identifying interference in a radar system of a motor vehicle, the method comprising:

receiving at least one incoming signal of the radar system;

determining detection information from the incoming signal;

performing an evaluation of the detection information by at least one neural network; and

using a result of the evaluation as a prognosis of interference with the incoming signal,

wherein the at least one neural network comprises at least one recurrent neural network which takes into account the evaluations of temporally preceding detection information, and wherein an output of the at least one neural network is used as the result of the evaluation, and

wherein the at least one neural network comprises at least one convolutional neural network, which receives the detection information as input, and the output of which is used as input to the at least one recurrent neural network.

2 . The method according to claim 1 , wherein performing the evaluation of the detection information comprises:

preprocessing the detection information of a detection cycle by max-pooling to reduce a data size of the detection information;

extracting information about the interference in the form of at least one interference in the incoming signal from the preprocessed detection information by the convolutional neural network; and

performing a prognosis of the at least one interference for a temporally subsequent detection cycle based on the extracted information or based on the evaluations of temporally preceding detection information.

3 . The method according to claim 1 , wherein using the result of the evaluation or using the output of the at least one neural network comprises providing the prognosis by an output of the frequency range in which the interference will be present in the future.

4 . The method according to claim 1 , wherein using the result of the evaluation or using the output of the at least one neural network comprises electronic outputting the result to an electronic system of the vehicle or a control unit of the vehicle.

5 . The method according to claim 1 , wherein the result of the evaluation comprises a segmentation of the detection information which indicates the predicted interference.

6 . The method according to claim 1 , wherein the at least one neural network is trained by:

storing a plurality of temporally successively determined detection information;

providing ground truth data by labeling of interferences in the detection information; and

training the at least one neural network using training data formed from the detection information and the ground truth data.

7 . The method according to claim 6 , wherein providing the ground truth data comprises manually labeling the interferences to teach the at least one neural network by the training to predict the interferences in the incoming signal.

8 . The method according to claim 2 , wherein for the detection cycle, a plurality of transmission signals of the radar system are transmitted in succession, each in at least one frequency range in order to receive an associated incoming signal, and wherein the transmission signal is implemented as at least one chirp with a time varying frequency within the frequency range.

9 . The method according to claim 8 , wherein determining the detection information is performed per detection cycle and comprises:

performing a mixing of the respective transmission signal and the associated incoming signal so as to obtain a baseband signal in each case; and

determining the detection information from the obtained baseband signals, the detection information being specific to an object detection in surroundings of the vehicle.

10 . The method according to claim 8 , wherein the result of the evaluation has an indication of a predicted interference frequency range in which the interference is predicted in a temporally subsequent detection cycle, wherein using the result of the evaluation comprises an automatic, at least partial adjustment of a frequency range in which the transmission signals are transmitted and which is at least partially outside the predicted interference frequency range, so that the frequency range is implemented as an at least partially variable frequency range.

11 . A radar system for detecting target objects in surroundings of a vehicle, the radar system comprising:

a processing device designed to carry out the steps:

determining detection information from an incoming signal of the radar system;

performing an evaluation of the detection information by at least one neural network; and

using a result of the evaluation as a prognosis of interference with the incoming signal,

wherein the at least one neural network comprises at least one recurrent neural network which takes into account the evaluations of temporally preceding detection information, and wherein an output of the at least one neural network is used as the result of the evaluation, and

wherein the at least one neural network comprises at least one convolutional neural network, which receives the detection information as input, and the output of which is used as input to the at least one recurrent neural network.

12 . A non-transitory computer-readable medium storing a computer program comprising instructions which, when executed by a processing device of a radar system, cause the processing device to perform the following steps:

determining detection information from an incoming signal of the radar system;

performing an evaluation of the detection information by at least one neural network; and

using a result of the evaluation as a prognosis of interference with the incoming signal,

wherein the at least one neural network comprises at least one recurrent neural network which takes into account the evaluations of temporally preceding detection information, and wherein an output of the at least one neural network is used as the result of the evaluation, and

wherein the at least one neural network comprises at least one convolutional neural network, which receives the detection information as input, and the output of which is used as input to the at least one recurrent neural network.

Assignments (1)

Priority Claims (1)

Continuity (2)

References Cited (10)