Robot and method for recognizing wake-up word thereof

1. A robot comprising:

a microphone configured to acquire a sound signal corresponding to a sound generated near the robot;

a camera;

an output interface including at least one of a display configured to output a wake-up screen or a speaker configured to output a wake-up sound when the robot wakes up;

an illumination sensor; and

a processor configured to:

recognize whether the acquired sound signal includes a voice of a person;

activate the camera based on recognizing that the acquired sound signal includes the voice of the person;

recognize whether the person is present in an image acquired by the activated camera;

set a wake-up word recognition sensitivity based on a result of recognizing whether a person is present in the acquired image;

measure, using the illumination sensor, an intensity of illumination of a space in which the robot is disposed;

increase the set wake-up word recognition sensitivity based on a sensing value of the illumination sensor being lower than a reference sensing value, and decrease the set wake-up word recognition sensitivity based on the sensing value of the illumination sensor being higher than the reference sensing value; and

recognize whether a wake-up word is included in voice data of a user acquired through the microphone based on the set wake-up word recognition sensitivity.

2. The robot of claim 1 , wherein the processor is configured to set the wake-up word recognition sensitivity to a first level when the person is recognized from the acquired image.

3. The robot of claim 2 , wherein the processor is configured to set the wake-up word recognition sensitivity to be higher than the first level when a number of people equal to or larger than a reference number are recognized from the acquired image.

4. The robot of claim 2 , wherein the processor is configured to set the wake-up word recognition sensitivity to a second level lower than the first level when no person is recognized from the acquired image.

5. The robot of claim 2 , wherein the processor is configured to measure a sound volume based on the sound signal if no person is recognized from the acquired image, and set the wake-up word recognition sensitivity based on the measured sound volume.

6. The robot of claim 5 , wherein the processor is configured to set the wake-up word recognition sensitivity to a second level lower than the first level when the measured sound volume is larger than a reference volume, and set the wake-up word recognition sensitivity to a third level higher than the first level when the measured sound volume is smaller than the reference volume.

7. The robot of claim 1 , wherein the processor is configured to recognize whether the voice of the person is included in the sound signal based on a signal component having a frequency band corresponding to the voice of the person, in the sound signal.

8. The robot of claim 7 , wherein the processor is configured to measure a sound volume based on the sound signal when it is recognized that the voice of the person is not included in the sound signal, and set the wake-up word recognition sensitivity to be lower as the measured sound volume is larger.

9. The robot of claim 1 , wherein the processor is configured to:

recognize at least one of a syllable, a character, or a word spoken by the user from the voice data,

compare the recognized at least one of the syllable, character, or word with the wake-up word, and

recognize whether the wake-up word is included in the voice data based on a comparison result and the wake-up word recognition sensitivity.

10. The robot of claim 9 , wherein the processor is configured to:

acquire a matching rate between the recognized at least one of the syllable, character, or word and the wake-up word, and

recognize that the voice data includes the wake-up word when the acquired matching rate is higher than a reference matching rate of a current set wake-up word recognition sensitivity.

11. The robot of claim 9 , further comprising a memory storing a learning model trained based on machine learning, and

wherein the processor is configured to recognize the at least one of the syllable, character, or word from feature points extracted from the voice data through the learning model.

12. The robot of claim 1 , wherein the processor is configured to control the output interface to output at least one of the wake-up screen or the wake-up sound as the robot wakes up when it is recognized that the voice data includes the wake-up word.

13. The robot of claim 1 , further comprising a transceiver configured to be connected to a server,

wherein the processor is configured to:

transmit the acquired sound signal to the server,

transmit the acquired voice data to the server,

receive a wake-up control signal based on a result of recognizing whether the wake-up word is included in the voice data from the server, and

control the output interface based on the received wake-up control signal.

14. A method of recognizing a wake-up word using a robot, the method comprising:

acquiring a sound signal through a microphone of the robot;

recognizing whether the acquired sound signal includes a voice of a person;

activating a camera included in the robot based on recognizing that the acquired sound signal includes the voice of the person;

recognizing whether a person is present in an image acquired by the activated camera;

setting a wake-up word recognition sensitivity based on a result of recognizing whether the person is present in the acquired image;

increasing or decreasing the set wake-up word recognition sensitivity based on a sensing value measured through an illumination sensor of the robot;

recognizing whether the wake-up word is included in voice data of a user acquired through the microphone based on the set wake-up word recognition sensitivity; and

waking up the robot based on recognizing that the wake-up word is included in the voice data.

15. The method of claim 14 , wherein the setting of the wake-up word recognition sensitivity comprises setting the wake-up word recognition sensitivity when the person is recognized from the acquired image to be higher than the wake-up word recognition sensitivity when no person is recognized.

16. The method of claim 14 , wherein

the setting of the wake-up word recognition sensitivity comprises:

measuring a sound volume based on the sound signal when no person is recognized from the acquired image; and

setting the wake-up word recognition sensitivity based on the measured sound volume.

17. The method of claim 16 , wherein

the setting of the wake-up word recognition sensitivity based on the measured sound volume comprises:

setting a wake-up word recognition sensitivity lower than the wake-up word recognition sensitivity of a case where the person is recognized when the sound volume is greater than a reference volume; and

setting a wake-up word recognition sensitivity higher than the wake-up word recognition sensitivity of the case where the person is recognized when the sound volume is smaller than the reference volume.

18. A server comprising:

a transceiver configured to receive a sound signal corresponding to a sound generated near a robot from the robot; and

a processor configured to:

receive an image in a vicinity of the robot from the robot through the transceiver when a voice of a person is recognized from the received sound signal;

recognize whether the person is present from the received image;

set a wake-up word recognition sensitivity based on a result of recognizing whether the person is present;

increase or decrease the set wake-up word recognition sensitivity based on a sensing value measured through an illumination sensor of the robot; and

recognize whether a wake-up word is included from voice data of a user received through the transceiver based on the set wake-up word recognition sensitivity.