High resolution radar simulation to train vehicle radar system neural network

Patent metadata

Field	Value
Publication number	US-12270891-B2
Application number	US-202217947298-A
Country	US
Kind code	B2
Filing date	Sep 19, 2022
Priority date	Sep 19, 2022
Publication date	Apr 8, 2025
Grant date	Apr 8, 2025

Abstract

A system includes a transmitter of a radar system to transmit transmitted signals, and a receiver of the radar system to receive received signals based on reflection of one or more of the transmitted signals by one or more objects. The system also includes a processor to train a neural network with reference data obtained by simulating a higher resolution radar system than the radar system to obtain a trained neural network. The trained neural network enhances detection of the one or more objects based on obtaining and processing the received signals in a vehicle. One or more operations of the vehicle are controlled based on the detection of the one or more objects.

First claim

What is claimed is:

1. A system comprising:
a transmitter of a radar system configured to transmit transmitted signals;
a receiver of the radar system configured to receive received signals based on reflection of one or more of the transmitted signals by one or more objects; and
a processor configured to:
obtain simulated received signals by using parameters of the radar system on reflection points;
obtain simulated high resolution received signals by using high resolution parameters for the high resolution radar system, the simulated high resolution received signals obtained by simulating a higher resolution radar system than the radar system; and
train a neural network by:
processing the simulated received signals to obtain a data cube indicating an intensity at a first set of ranges, a first set of hypotheses of Doppler frequency, and a first set of angle hypotheses; and
processing the simulated high resolution received signals to obtain a high resolution data cube indicating an intensity at a second set of ranges, a second set of hypotheses of Doppler frequency, and a second set of angle hypotheses, a number of the second set of ranges, the second set of hypotheses of Doppler frequency, and the second set of angle hypotheses being greater than a number of the first set of ranges, the first set of hypotheses of Doppler frequency, and the first set of angle hypotheses;

wherein the trained neural network enhances detection of the one or more objects based on obtaining and processing the received signals in a vehicle and one or more operations of the vehicle are controlled based on the detection of the one or more objects.

2. The system according to claim 1 , wherein the processor is configured to simulate the higher resolution radar system to include more antennas or more closely spaced antennas that span a wider aperture than the radar system.

3. The system according to claim 1 , wherein the processor is configured to simulate the higher resolution radar system to include more of the transmitted signals or more closely spaced transmitted signals than the radar system.

4. The system according to claim 1 , wherein the processor is configured to simulate the higher resolution radar system to include a higher bandwidth than the radar system.

5. The system according to claim 1 , wherein the processor is configured to train the neural network by obtaining the reflection points from a lidar system.

6. The system according to claim 1 , wherein the processor is configured to train the neural network by generating the reflection points.

7. The system according to claim 1 , wherein the processor is configured to train the neural network by providing the data cube to the neural network and obtaining a neural network output.

8. The system according to claim 7 , wherein the processor is configured to train the neural network by obtaining a metric that indicates a match between the neural network output and the high resolution data cube and updating parameters of the neural network based on the metric.

9. The system of claim 1 , wherein the transmitted signals are linear frequency modulated continuous wave signals.

10. The system of claim 1 , wherein the neural network is a convolutional neural network.

11. A non-transitory computer-readable medium configured to store instructions that, when processed by one or more processors, cause the one or more processors to implement a method, the method comprising:
obtaining received signals resulting from reflection of one or more transmitted signals, transmitted by a transmitter of a radar system, by one or more objects and reception of the received signals by a receiver of the radar system; and
obtaining simulated received signals by using parameters of the radar system on reflection points;
obtaining simulated high resolution received signals by using high resolution parameters for a high resolution radar system that simulates a higher resolution radar system than the radar system; and
training the neural network by:
processing the simulated received signals to obtain a data cube indicating an intensity at a first set of ranges, a first set of hypotheses of Doppler frequency, and a first set of angle hypotheses; and
processing the simulated high resolution received signals to obtain a high resolution data cube indicating an intensity at a second set of ranges, a second set of hypotheses of Doppler frequency, and a second set of angle hypotheses, a number of the second set of ranges, the second set of hypotheses of Doppler frequency, and the second set of angle hypotheses being greater than a number of the first set of ranges, the first set of hypotheses of Doppler frequency, and the first set of angle hypotheses;

wherein the trained neural network enhances detection of the one or more objects based on obtaining and processing the received signals in a vehicle and one or more operations of the vehicle are controlled based on the detection of the one or more objects.

12. The non-transitory computer-readable medium according to claim 11 , wherein the method also includes simulating the higher resolution radar system to include more antennas or more closely spaced antennas that span a wider aperture than the radar system.

13. The non-transitory computer-readable medium according to claim 11 , wherein the method also includes simulating the higher resolution radar system to include more of the transmitted signals or more closely spaced transmitted signals than the radar system.

14. The non-transitory computer-readable medium according to claim 11 , wherein the method also includes simulating the higher resolution radar system to include a higher bandwidth than the radar system.

15. The non-transitory computer-readable medium according to claim 11 , wherein the method also includes training the neural network by obtaining the reflection points from a lidar system.

16. The non-transitory computer-readable medium according to claim 11 , wherein the method also includes training the neural network by generating the reflection points.

17. The non-transitory computer-readable medium according to claim 11 , wherein the method also includes training the neural network by providing the data cube to the neural network and obtaining a neural network output.

18. The non-transitory computer-readable medium according to claim 17 , wherein the method also includes training the neural network by obtaining a metric that indicates a match between the neural network output and the high resolution data cube and updating parameters of the neural network based on the metric.

19. The non-transitory computer-readable medium of claim 11 , wherein the transmitted signals are linear frequency modulated continuous wave signals.

20. The non-transitory computer-readable medium of claim 11 , wherein the neural network is a convolutional neural network.

Assignees

• Gm Global Tech Operations Llc

Inventors

• Bialer Oded
• Haitman Yuval
• Levi Dan

Classifications

• G01S13/505

  using Doppler effect for determining closest range to a target or corresponding time, e.g. miss-distance indicator · CPC title

• G01S7/4095

  the external reference signals being modulated, e.g. rotating a dihedral reflector or modulating a transponder for simulation of a Doppler echo · CPC title

• G01S7/417 Primary

  involving the use of neural networks · CPC title

• G01S13/584

  adapted for simultaneous range and velocity measurements · CPC title

• G06N3/08

  Learning methods · CPC title