Driver passenger detection using ultrasonic sensing

1. A computer-implemented method comprising:

receiving, by a computing device and from a signal transmitter in a vehicle, ultrasonic sensing data associated with a plurality of mobile devices in the vehicle, wherein the plurality of mobile devices are associated with a plurality of occupants in the vehicle, and wherein the signal transmitter comprises a high-frequency ultrasonic transmitter;

determining, by the computing device, a unique identifier associated with each of the plurality of mobile devices;

determining, based on the ultrasonic sensing data and the unique identifier, a relative distance from the signal transmitter to each of the plurality of mobile devices in the vehicle;

determining that a particular occupant, from the plurality of occupants in the vehicle, is a driver in the vehicle by comparing the relative distance from the signal transmitter to each of the plurality of mobile devices; and

storing, in a sensing database, a frequency signature associated with a particular mobile device of the plurality of mobile devices, wherein the frequency signature indicates a mapping between an identifier of the particular mobile device and a particular frequency that the signal transmitter used to communicate with the particular mobile device.

2. The method of claim 1 , further comprising:

receiving, by the computing device, sensor data from a sensor array associated with the vehicle; and

determining, using a machine learning classifier and based on the ultrasonic sensing data and the sensor data, that the particular occupant is the driver of the vehicle.

3. The method of claim 2 , wherein the sensor array comprises an infrared sensor, a sound sensor, a pressure sensor and a motion sensor.

4. The method of claim 2 , wherein the machine learning classifier comprise a supervised machine learning classifier and an unsupervised machine learning classifier.

5. The method of claim 2 , wherein determining, using the machine learning classifier, that the particular occupant is the driver of the vehicle comprises:

prior to using the machine learning classifier, training, using training data comprising predefined labels associated with a set of occupants in the vehicle, the machine learning classifier to output predicted labels for occupants associated with the vehicle;

providing, as input to the trained machine learning classifier, the sensor data and the ultrasonic sensing data; and

receiving, as output from the trained machine learning classifier and based on the sensor data and the ultrasonic sensing data, a role label of the particular occupant indicating whether the particular occupant is the driver of the vehicle.

6. The method of claim 5 , wherein training the machine learning classifier comprises:

generating, using the machine learning classifier, a first set of predicted labels;

determining that the first set of predicted labels have corresponding confidence scores falling below a threshold value; and

regenerating, using the machine learning classifier, a second set of predicted labels, wherein the second set of predicted labels having confidence scores above the threshold value.

7. The method of claim 1 , further comprising:

receiving, from the signal transmitter, the ultrasonic sensing data associated with a plurality of frequencies, wherein each frequency corresponds to one of the plurality of the mobile devices in the vehicle.

8. The method of claim 7 , wherein each frequency corresponds to a communication channel between the signal transmitter and the corresponding mobile device.

9. A computing device comprising:

one or more processors; and

memory storing instructions that, when executed by the one or more processors, cause the computing device to:

receive, from a signal transmitter in a vehicle, ultrasonic sensing data associated with a plurality of mobile devices in the vehicle, wherein the plurality of mobile devices are associated with a plurality of occupants in the vehicle, and wherein the signal transmitter comprises a high-frequency ultrasonic transmitter;

determine a unique identifier associated with each of the plurality of mobile devices;

determine, based on the ultrasonic sensing data and the unique identifier, a relative distance from the signal transmitter to each of the plurality of mobile devices in the vehicle;

determine that a particular occupant, from the plurality of occupants in the vehicle, is a driver in the vehicle by comparing the relative distance from the signal transmitter to each of the plurality of mobile devices; and

store, in a sensing database, a frequency signature associated with a particular mobile device of the plurality of mobile devices, wherein the frequency signature indicates a mapping between an identifier of the particular mobile device and a particular frequency that the signal transmitter used to communicate with the particular mobile device.

10. The computing device of claim 9 , wherein the instructions, when executed by the one or more processors, cause the computing device to:

receive sensor data from a sensor array associated with the vehicle; and

determine, using a machine learning classifier and based on the ultrasonic sensing data and the sensor data, that the particular occupant is the driver of the vehicle.

11. The computing device of claim 10 , wherein the sensor array comprises an infrared sensor, a sound sensor, a pressure sensor and a motion sensor.

12. The computing device of claim 10 , wherein the machine learning classifier comprises a supervised machine learning classifier and an unsupervised machine learning classifier.

13. The computing device of claim 10 , wherein the instructions, when executed by the one or more processors, cause the computing device to:

prior to using the machine learning classifier, training, using training data comprising predefined labels associated with a set of occupants in the vehicle, the machine learning classifier to output predicted labels for occupants associated with the vehicle;

provide, as input to the trained machine learning classifier, the sensor data and the ultrasonic sensing data; and

receive, as output from the trained machine learning classifier and based on the sensor data and the ultrasonic sensing data, a role label of the particular occupant indicating whether the particular occupant is the driver of the vehicle.

14. The computing device of claim 10 , wherein the instructions, when executed by the one or more processors, cause the computing device to:

generate, using the machine learning classifier, a first set of predicted labels;

determine that the first set of predicted labels have corresponding confidence scores falling below a threshold value; and

regenerate, using the machine learning classifier, a second set of predicted labels, wherein the second set of predicted labels having confidence scores above the threshold value.

15. A non-transitory computer-readable medium storing instructions that, when executed, configure a computing device to:

receive, from a signal transmitter in a vehicle, ultrasonic sensing data associated with a plurality of mobile devices in the vehicle, wherein the plurality of mobile devices are associated with a plurality of occupants in the vehicle, and wherein the signal transmitter comprises a high-frequency ultrasonic transmitter;

determine a unique identifier associated with each of the plurality of mobile devices;

determine, based on the ultrasonic sensing data and the unique identifier, a relative distance from the signal transmitter to each of the plurality of mobile devices in the vehicle;

determine that a particular occupant, from the plurality of occupants in the vehicle, is a driver in the vehicle by comparing the relative distance from the signal transmitter to each of the plurality of mobile devices; and

store, in a sensing database, a frequency signature associated with a particular mobile device of the plurality of mobile devices, wherein the frequency signature indicates a mapping between an identifier of the particular mobile device and a particular frequency that the signal transmitter used to communicate with the particular mobile device.

16. The non-transitory computer-readable medium of claim 15 , wherein the instructions, when executed, configure the computing device to:

receive sensor data from a sensor array associated with the vehicle; and

determine, using a machine learning classifier and based on the ultrasonic sensing data and the sensor data, that the particular occupant is the driver of the vehicle.

17. The non-transitory computer-readable medium of claim 16 , wherein the sensor array comprises an infrared sensor, a sound sensor, a pressure sensor and a motion sensor.

18. The non-transitory computer-readable medium of claim 16 , wherein the instructions, when executed, configure the computing device to:

prior to using the machine learning classifier, training, using training data comprising predefined labels associated with a set of occupants in the vehicle, the machine learning classifier to output predicted labels for occupants associated with the vehicle;

provide, as input to the trained machine learning classifier, the sensor data and the ultrasonic sensing data; and

receive, as output from the trained machine learning classifier and based on the sensor data and the ultrasonic sensing data, a role label of the particular occupant indicating whether the particular occupant is the driver of the vehicle.