Multi-sensor based user interface

• US 10,509,479 B2
• Filed: 07/24/2018
• Issued: 12/17/2019
• Est. Priority Date: 03/03/2015
• Status: Active Grant

• Current Assignee: NVIDIA Corporation
• Original Assignee: NVIDIA Corporation
• Inventors: Molchanov, Pavlo, Gupta, Shalini, Kim, Kihwan, Pulli, Kari
• Primary Examiner(s): Sitta, Grant
• Application Number:
  US16/044,239
• Publication Number:
  US 20180341333A1
• Time in Patent Office: 511 Days
• CPC Class Codes: B60R 11/04 Mounting of cameras operative during drive; Arrangement of controls thereof relative to the vehicle
• CPC Class Codes: G01S 13/42 Simultaneous measurement of distance and other co-ordinates indirect measurement G01S13/46
• CPC Class Codes: G01S 13/4454 phase comparisons monopulse, i.e. comparing the echo signals received by an interferometric antenna arrangement
• CPC Class Codes: G01S 13/584 adapted for simultaneous range and velocity measurements
• CPC Class Codes: G01S 7/0233 Avoidance by phase multiplex
• CPC Class Codes: G01S 7/0234 Avoidance by code multiplex
• CPC Class Codes: G01S 7/352 Receivers
• CPC Class Codes: G01S 7/415 Identification of targets based on measurements of movement associated with the target
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06F 18/24 Classification techniques
• CPC Class Codes: G06F 18/25 Fusion techniques
• CPC Class Codes: G06F 3/011 Arrangements for interaction with the human body, e.g. for user immersion in virtual reality blind teaching G09B21/00
• CPC Class Codes: G06F 3/017 Gesture based interaction, e.g. based on a set of recognized hand gestures interaction based on gestures traced on a digitiser G06F3/04883
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 3/082 modifying the architecture, e.g. adding, deleting or silencing nodes or connections
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: G06V 40/28 Recognition of hand or arm movements, e.g. recognition of deaf sign language static hand signs G06V40/113

Abstract

An apparatus and method for gesture detection and recognition. The apparatus includes a processing element, a radar sensor, a depth sensor, and an optical sensor. The radar sensor, the depth sensor, and the optical sensor are coupled to the processing element, and the radar sensor, the depth sensor, and the optical sensor are configured for short range gesture detection and recognition. The processing element is further configured to detect and recognize a hand gesture based on data acquired with the radar sensor, the depth sensor, and the optical sensor.

Claims

1. An apparatus for gesture detection and recognition, the apparatus comprising:

a processing element;

a radar sensor;

a depth sensor; and

an optical sensor, wherein the radar sensor, the depth sensor, and the optical sensor are coupled to the processing element, wherein the processing element is configured to identify a type of gesture, wherein the data acquired with the radar sensor is registered to the data acquired with the depth sensor to produce registration information for the radar sensor and the depth sensor, wherein the registration information is used with calibration information for the optical sensor and the depth sensor to produce a velocity image comprising a plurality of pixels and a velocity value per pixel in the plurality of pixels, and wherein velocity values of the velocity image are used by the processing element to classify the gesture.

2. The apparatus as described in claim 1, wherein the radar sensor is in an always-on mode during a period in which the depth sensor and the optical sensor are turned off, wherein the depth sensor and the optical sensor are activated and a gesture recognition process to identify the type of the gesture is performed only in response to the radar sensor detecting an amount of motion above a non-zero threshold amount that lasts for at least a non-zero threshold length of time.

3. The apparatus as described in claim 2, wherein the threshold amount of motion is selected from the group consisting of:

a threshold velocity of the motion; and

a threshold distance of the motion.

4. The apparatus as described in claim 1, wherein the radar sensor, the depth sensor, and the optical sensor are a portion of a user interface device of a vehicle.

5. The apparatus as described in claim 1, wherein a portion of the processing element is configured to function as a deep neural network (DNN).

6. The apparatus as described in claim 5, wherein the DNN comprises two 3D convolutional layers and two fully-connected layers.

7. The apparatus as described in claim 1, wherein the radar sensor, the depth sensor, and the optical sensor are configured for gesture detection and recognition under low light conditions.

8. The apparatus as described in claim 1, wherein the processing element is a graphics processing unit (GPU).

9. The apparatus as described in claim 1, wherein the radar sensor, the depth sensor, and the optical sensor are configured for gesture detection and recognition within a range of one meter.

10. The apparatus as described in claim 1, wherein the gesture is a dynamic hand gesture and wherein further the processing element is configured to automatically determine a command associated with the dynamic hand gesture.

11. A system for hand gesture recognition, the system comprising:

a processor;

a first sensor comprising a radar;

a second sensor comprising a depth sensor; and

a third sensor comprising an optical sensor, wherein the first sensor, the second sensor, and third sensor are coupled to the processor, wherein further the processor is configured to identify a type of gesture of a hand by combining data acquired with the first sensor, data acquired with the second sensor, and data acquired with the third sensor, wherein the data acquired with the radar is registered to the data acquired with the depth sensor to produce registration information for the radar sensor and the depth sensor, wherein registering the data acquired with the radar to the data acquired with the depth sensor comprises transforming three-dimension (3D) coordinates of the data acquired with the radar to the depth sensor'"'"'s coordinate frame, wherein the registration information is used with calibration information for the optical sensor and the depth sensor to produce a velocity image comprising a velocity value per pixel in the velocity image, and wherein velocity values for a portion of the velocity image corresponding to the hand are used by the processing element to identify the type of the gesture.

12. The system as described in claim 11, wherein the radar sensor is in an always-on mode during a period in which the depth sensor and the optical sensor are turned off, wherein the depth sensor and the optical sensor are activated and a gesture recognition process to identify the type of the gesture is performed only in response to the radar sensor detecting an amount of motion above a threshold amount that lasts for at least a threshold length of time.

13. The system as described in claim 12, wherein the threshold amount of motion is selected from the group consisting of:

a threshold velocity of the motion; and

a threshold distance of the motion.

14. The system as described in claim 11, wherein the first sensor, the second sensor, and the third sensor are a portion of a user interface device for use in a vehicle.

15. The system as described in claim 11, wherein a portion of the processor is configured to function as a deep neural network (DNN).

16. The system as described in claim 11, wherein the processor is a graphics processing unit (GPU).

17. The system as described in claim 11, wherein hand gesture is a dynamic hand gesture and wherein further the processing element is configured to automatically determine a command associated with the dynamic hand gesture.

18. A method for hand gesture recognition, the method comprising:

receiving, at a processor, data from a first sensor comprising a radar;

receiving, at the processor, data from a second sensor comprising a depth sensor;

receiving, at the processor, data from a third sensor comprising an optical sensor;

the processor combining the data from the first sensor, the data from the second sensor, and the data from the third sensor, wherein the data acquired with the radar is registered to the data acquired with the depth sensor to produce registration information for the radar sensor and the depth sensor;

the processor using the registration information with calibration information for the optical sensor and the depth sensor to produce a velocity image comprising a plurality of pixels and a velocity value per pixel in the plurality of pixels; and

the processor identifying a type of gesture of a hand using velocity values for a portion of the velocity image corresponding to the hand.

19. The method of claim 18, wherein registering the data acquired with the radar to the data acquired with the depth sensor comprises transforming three-dimension (3D) coordinates of the data acquired with the radar to the depth sensor'"'"'s coordinate frame.

20. The method of claim 18, further comprising the processor determining a command associated with the gesture.