Radio frequency (RF) object detection using radar and machine learning

Patent metadata

Field	Value
Publication number	US-11391817-B2
Application number	US-201916405793-A
Country	US
Kind code	B2
Filing date	May 7, 2019
Priority date	May 11, 2018
Publication date	Jul 19, 2022
Grant date	Jul 19, 2022

Abstract

Embodiments described herein can address these and other issues by using radar machine learning to address the radio frequency (RF) to perform object identification, including facial recognition. In particular, embodiments may obtain IQ samples by transmitting and receiving a plurality of data packets with a respective plurality of transmitter antenna elements and receiver antenna elements, where each data packet of the plurality of data packets comprises one or more complementary pairs of Golay sequences. I/Q samples indicative of a channel impulse responses of an identification region obtained from the transmission and reception of the plurality of data packets may then be used to identify, with a random forest model, a physical object in the identification region.

First claim

What is claimed is:

1. A method of performing object recognition using radio frequency (RF) signals at an electronic device, the method including:
obtaining a set of I/Q samples by wirelessly transmitting, with the electronic device within a scanning period, a plurality of data packets with a plurality of transmitter antenna elements and receiving the plurality of data packets at a plurality of receiver antenna elements, such that each receiver antenna element of the plurality of receiver antenna elements receives at least one data packet from each transmitter antenna element of the plurality of transmitter antenna elements, wherein:
each data packet of the plurality of data packets comprises a complementary pair of Golay sequences, and each I/Q sample is indicative of a channel impulse response of at least a portion of an identification region, the channel impulse response obtained from a packet transmitted by a transmitter antenna element and received by a receiver antenna element;
selecting, with a processing unit of the electronic device, values from a sample subset from the set of I/Q samples, wherein selecting the values from the sample subset includes performing Principal Component Analysis (PCA) on the set of I/Q samples and choosing the sample subset from the set of I/Q samples based on a result of the PCA; and
using, with the processing unit of the electronic device, a random forest model on the selected values from the sample subset to identify a physical object in the identification region.

2. The method of claim 1 , wherein the physical object comprises a face.

3. The method of claim 1 , wherein selecting the sample subset is based on a physical distance, from the electronic device, sampled by the sample subset.

4. The method of claim 1 , wherein selecting the sample subset comprises downsampling the set of I/Q samples.

5. The method of claim 1 , wherein using the random forest model includes using the random forest model on different subsets of the set of I/Q samples, and identifying the physical object is based on an output of the random forest model for each of the different subsets of the set of I/Q samples.

6. The method of claim 1 , wherein the random forest model comprises a multi-class classifier trained using positive and negative RF images.

7. The method of claim 1 , wherein the plurality of data packets are wirelessly transmitted by the plurality of transmitter antenna elements using RF signals within a range of 57-64 GHz.

8. The method of claim 1 , wherein the scanning period is 100 ms or less.

9. The method of claim 1 , wherein each packet of the plurality of data packets is transmitted in 6 μs or less.

10. The method of claim 1 , further comprising adjusting a length of the scanning period to accommodate a time during which the electronic device is used for data communication.

11. An electronic device comprising:
a plurality of transmitter antenna elements;
a plurality of receiver antenna elements;
a memory; and
a processing unit communicatively coupled with the plurality of transmitter antenna elements, the plurality of receiver antenna elements, and the memory, wherein the processing unit is configured to:
obtain a set of I/Q samples by wirelessly transmitting, within a scanning period, a plurality of data packets with the plurality of transmitter antenna elements and receiving the plurality of data packets at the plurality of receiver antenna elements, such that each receiver antenna element of the plurality of receiver antenna elements receives at least one data packet from each transmitter antenna element of the plurality of transmitter antenna elements, wherein:
each data packet of the plurality of data packets comprises a complementary pair of Golay sequences, and
each I/Q sample is indicative of a channel impulse response of at least a portion of an identification region, the channel impulse response obtained from a packet transmitted by a transmitter antenna element and received by a receiver antenna element;

select from a sample subset from the set of I/Q samples, wherein the processing unit is configured to select the values from the sample subset at least in part by performing Principal Component Analysis (PCA) on the set of I/Q samples and choosing the sample subset from the set of I/Q samples based on a result of the PCA; and
use a random forest model on the selected values from the sample subset to identify a physical object in the identification region.

12. The electronic device of claim 11 , wherein the electronic device is configured to identify physical objects comprising faces.

13. The electronic device of claim 11 , wherein the processing unit is configured to select the sample subset based on a physical distance, from the electronic device, sampled by the sample subset.

14. The electronic device of claim 11 , wherein the processing unit is configured to select the sample subset at least in part by downsampling the set of I/Q samples.

15. The electronic device of claim 11 , wherein the processing unit is configured to use the random forest model at least in part by using the random forest model on different subsets of the set of I/Q samples, and wherein the processing unit is further configured to identify the physical object based on an output of the random forest model for each of the different subsets of the set of I/Q samples.

16. The electronic device of claim 11 , wherein the random forest model comprises a multi-class classifier trained using positive and negative RF images.

17. The electronic device of claim 11 , wherein plurality of transmitter antenna elements are configured to wirelessly transmit the plurality of data packets using RF signals within a range of 57-64 GHz.

18. The electronic device of claim 11 , wherein the electronic device is configured to transmit the plurality of data packets such that the scanning period is 100 ms or less.

19. The electronic device of claim 11 , wherein the electronic device is configured to transmit the plurality of data packets such that each packet of the plurality of data packets is transmitted in 6 μs or less.

20. The electronic device of claim 11 , wherein the processing unit is further configured to adjust a length of the scanning period to accommodate a time during which the electronic device is used for data communication.

21. A device comprising:
means for obtaining a set of I/Q samples by wirelessly transmitting, within a scanning period, a plurality of data packets with a plurality of transmitter antenna elements and receiving the plurality of data packets at a plurality of receiver antenna elements, such that each receiver antenna element of the plurality of receiver antenna elements receives at least one data packet from each transmitter antenna element of the plurality of transmitter antenna elements, wherein:
each data packet of the plurality of data packets comprises a complementary pair of Golay sequences, and
each I/Q sample is indicative of a channel impulse response of at least a portion of an identification region, the channel impulse response obtained from a packet transmitted by a transmitter antenna element and received by a receiver antenna element;

means for selecting values from a sample subset from the set of I/Q samples, wherein the means for selecting the values from the sample subset includes means for performing Principal Component Analysis (PCA) on the set of I/Q samples and choosing the sample subset from the set of I/Q samples based on a result of the PCA;

Assignees

• Qualcomm Inc

Inventors

• Sambhwani Sharad
• Sanderovich Amichai
• Hemo Evyatar
• Levitan Evgeny
• Hof Eran
• Salama Mohammad Faroq
• Sarkis Michel Adib
• Bi Ning
• Qi Yingyong

Classifications

• G06V10/7715

  Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods · CPC title

• G06V10/143

  Sensing or illuminating at different wavelengths · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G01S7/41 Primary

  using analysis of echo signal for target characterisation; Target signature; Target cross-section · CPC title

• G01S7/352 Primary

  Receivers · CPC title