Training data for a motion detection system using data from a sensor device

• US 10,318,890 B1
• Filed: 05/23/2018
• Issued: 06/11/2019
• Est. Priority Date: 05/23/2018
• Status: Active Grant

• Current Assignee: Cognitive Systems Corp.
• Original Assignee: Cognitive Systems Corp.
• Inventors: Kravets, Oleksiy, Ravkine, Mikhail
• Primary Examiner(s): Misir, Dave
• Application Number:
  US15/987,805
• Time in Patent Office: 384 Days
• Field of Search: 706 12, 706 20
• CPC Class Codes: G01S 13/003 Bistatic radar systems; Multistatic radar systems
• CPC Class Codes: G01S 13/886 for alarm systems alarms with electrical actuation G08B13/22
• CPC Class Codes: G01S 7/415 Identification of targets based on measurements of movement associated with the target
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: H04L 25/0224 using sounding signals
• CPC Class Codes: H04W 24/08 Testing, supervising or monitoring using real traffic
• CPC Class Codes: H04W 4/029 Location-based management or tracking services
• CPC Class Codes: H04W 4/38 for collecting sensor information

Abstract

In a general aspect of the examples described, sensor data from a sensor device are used to improve training data for a motion detection system. In some aspects, “ground truth” labels, sensor data, and channel information are received for a motion detection training period. The labels and sensor data are analyzed to detect variances between the labels and sensor data. The labels are corrected or augmented based on the sensor data. The channel information is tagged with the labels and provided as training information to train the motion detection system.

Claims

1. A motion detection training method comprising:

receiving, by one or more processors of a data processing apparatus, channel information based on training signals, the training signals comprising wireless signals transmitted by a wireless communication device through a space between the wireless communication device and a second wireless communication device during a motion detection training time period supervised by a user;

receiving, by the one or more processors of the data processing apparatus, a label comprising a first set of motion information input to a user interface by the user of a mobile device in the space during the motion detection training time period;

receiving, by the one or more processors of the data processing apparatus, sensor data comprising a second set of motion information collected by a wearable sensor device of the user in the space during the motion detection training time period;

analyzing, by the one or more processors of the data processing apparatus, the label and the sensor data to detect a variance between the label and the sensor data for the motion detection training time period, wherein analyzing the label and the sensor data comprises;

detecting the variance when the analysis indicates that the sensor data does not correspond to the label; and

either correcting the label based on the received sensor data, or augmenting the label with the received sensor data, depending on the variance detected;

generating, by the one or more processors of the data processing apparatus, a tag for the channel information based on the analysis of the label and the sensor data; and

providing, by the one or more processors of the data processing apparatus, the channel information and the tag as training input to train a motion detection system.

2. The motion detection training method of claim 1, wherein generating the tag based on the analysis of the label and the sensor data improves accuracy of the tag compared to generating the tag only from the label.

3. The motion detection training method of claim 1, wherein the sensor device comprises at least one of:

an accelerometer;

a gyroscope;

a compass;

a pedometer; and

a barometer.

4. The motion detection training method of claim 1, wherein the mobile device comprises the sensor device.

5. The motion detection training method of claim 1, wherein the variance between the label and the sensor data comprises at least one of:

a difference between a timestamp in the label and a timestamp in the sensor data;

a difference between a type of measurement indicated in the label and a type of measurement in the sensor data;

or

additional information provided in the sensor data but not the label.

6. The motion detection training method of claim 1, wherein providing the channel information and the tag as training input to train the motion detection system comprises providing a tagged channel information to a machine learning system.

7. The motion detection training method of claim 6, wherein the tag indicates at least one of:

whether motion occurred in the space during the motion detection training time period;

a direction of motion that occurred in the space during the motion detection training time period;

a location of motion that occurred in the space during the motion detecting training time period;

or

a time at which motion occurred in the space during the motion detection training time period.

8. A motion detection training system comprising:

one or more processors;

memory storing instructions that, when executed by the one or more processors, cause the one or more processors to;

receive channel information based on training signals, the training signals comprising wireless signals transmitted by a wireless communication device to a second wireless communication device through a space during a motion detection training time period supervised by a user;

receive a label comprising a first set of motion information input to a user interface by the user of a mobile device in the space during the motion detection training time period;

receive sensor data comprising a second set of motion information collected by a wearable sensor device of the user in the space during the motion detection training time period;

wherein analyzing the label and the sensor data comprises;

detecting the variance when the analysis indicates that the sensor data does not correspond to the label; and

either correcting the label based on the received sensor data, or augmenting the label with the received sensor data, depending on the variance detected;

analyze the label and the sensor data to detect a variance between the label and the sensor data for the motion detection training time period;

generate a tag for the channel information based on the analysis of the label and the sensor data; and

provide the channel information and the tag as training input to train a motion detection system.

9. The motion detection training system of claim 8, wherein generating the tag based on the analysis of the label and the sensor data improves accuracy of the tag compared to generating the tag only from the label.

10. The motion detection training system of claim 8, wherein the sensor device comprises at least one of:

an accelerometer;

a gyroscope;

a compass;

a pedometer; and

a barometer.

11. The motion detection training system of claim 8, wherein the mobile device comprises the sensor device.

12. The motion detection training system of claim 8, wherein the variance between the label and the sensor data comprises at least one of:

a difference between a timestamp in the label and a timestamp in the sensor data;

a difference between a type of measurement indicated in the label and a type of measurement in the sensor data;

or

additional information provided in the sensor data but not the label.

13. The motion detection training system of claim 8, wherein providing the channel information and the tag as training input to train the motion detection system comprises providing a tagged channel information to a machine learning system.

14. The motion detection training system of claim 13, wherein the tag indicates at least one of:

whether motion occurred in the space during the motion detection training time period;

a direction of motion that occurred in the space during the motion detection training time period;

a location of motion that occurred in the space during the motion detection training time period;

or

a time at which motion occurred in the space during the motion detection training time period.

15. A non-transitory computer readable medium comprising instructions that when executed by data processing apparatus cause the data processing apparatus to perform operations comprising:

receiving channel information based on training signals, the training signals comprising wireless signals transmitted by a wireless communication device to a second wireless communication device through a space during a motion detection training time period supervised by a user;

receiving a label comprising a first set of motion information input to a user interface by the user of a mobile device in the space during the motion detection training time period;

receiving sensor data comprising a second set of motion information collected by a wearable sensor device of the user in the space during the motion detection training time period;

analyzing the label and the sensor data to detect a variance between the label and the sensor data for the motion detection training time period, wherein analyzing the label and the sensor data comprises;

detecting the variance when the analysis indicates that the sensor data does not correspond to the label; and

either correcting the label based on the received sensor data, or augmenting the label with the received sensor data, depending on the variance detected;

generating a tag for the channel information based on the analysis of the label and the sensor data; and

providing the channel information and the tag as training input to train a motion detection system.

16. The computer readable medium of claim 15, wherein generating the tag based on the analysis of the label and the sensor data improves accuracy of the tag compared to generating the tag only from the label.

17. The computer readable medium of claim 15, wherein the sensor device comprises at least one of:

an accelerometer;

a gyroscope;

a compass;

a pedometer; and

a barometer.

18. The computer readable medium of claim 15, wherein the mobile device comprises the sensor device.

19. The computer readable medium of claim 15, wherein the variance between the label and the sensor data comprises at least one of:

a difference between a timestamp in the label and a timestamp in the sensor data;

a difference between a type of measurement indicated in the label and a type of measurement in the sensor data;

or

additional information provided in the sensor data but not the label.

20. The computer readable medium of claim 15, wherein providing the channel information and the tag as training input to train the motion detection system comprises providing a tagged channel information to a machine learning system.

21. The computer readable medium of claim 20, wherein the tag indicates at least one of:

whether motion occurred in the space during the motion detection training time period;

a direction of motion that occurred in the space during the motion detection training time period;

a location of motion that occurred in the space during the motion detection training time period;

or

a time at which motion occurred in the space during the motion detection training time period.