JP2008087140A - Speech recognition robot and control method of speech recognition robot - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a speech recognition robot which responds to a speaker always in a state of facing the direction of the speaker making a speech and also to provide its control method. <P>SOLUTION: The speech recognition robot is furnished with: a sound source specifying part to specify a generating direction of speech; a speech recognizing part to receive the speech and to recognize its contents; an image pick-up part to acquire a picked-up image as an image data concerning more than two directions; a discriminating part to discriminate a position and a direction of a detected face; and an extracting part to extract the detected face. The generating direction of the speech specified by the speech specifying part is picked up by the image pick-up part, a face of a figure existing in the image picked up by the image pick-up part is detected by a face detecting part, a face turning to the direction of the robot is discriminated out of the detected faces by the discriminating part, the face in a position nearest from the robot is extracted by the extracting part, a direction to pick up the image of the image pick-up part is changed in accordance with the extracted face, and the extracted faces are continuously picked up. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a speech-recognition type speech recognition robot that recognizes the content of speech obtained by speaking to a person and responds to the spoken person, and a control method for such a speech recognition robot.

  In recent years, it has become a conversational type for conversing with humans by acquiring the content spoken by humans (speakers) as speech data, recognizing the content and speaking the response sentence corresponding to the content as speech. Speech recognition systems are being developed. Such a speech recognition system stores a large amount of speech data to be uttered as a response sentence in a storage area inside the system, selects speech data most related to the content of the recognized speech, and selects the selected speech data as speech. Speak. Such a speech recognition system acquires the spoken content as speech data, and divides the speech data into clauses of a predetermined language, so that the response most suitable for the content is based on the order and proximity of each clause. Select a sentence. (For example, Patent Document 1)

JP 2004-109323 A

  By the way, the voice recognition system as described above can identify the direction in which the voice is uttered by receiving the voice using a directional microphone or the like when distinguishing the acquired voice from the speaker. Therefore, when such a speech recognition system is incorporated into a human interactive robot (hereinafter referred to as a “speech recognition robot”), the speaker is moved by moving the speech recognition robot so as to face a specified direction. When receiving a voice from the speaker, the robot can be seen as if it is having a conversation by recognizing the speaker as a conversation target.

  However, if there are multiple speakers and these speakers are moving relative to the robot, the direction of the speaker viewed from the robot cannot be specified, and the robot sees the speaker. The direction may not be determined.

  The present invention has been made to solve such a problem, and is a speech recognition robot capable of responding to a speaker in a state in which the speaker is always directed to the speaker, and a method for controlling the speech recognition robot. The purpose is to provide.

  A voice recognition robot according to the present invention includes a sound source identifying unit that identifies a direction in which a voice is generated, a voice recognition unit that receives the generated voice and recognizes the contents thereof, and images at least two directions, An imaging unit that acquires a captured image as image data, a face detection unit that detects a human face existing in the captured image, a determination unit that determines the position and orientation of the detected face, and a detected face An image capturing unit that captures the direction in which the sound specified by the sound source specifying unit is generated, and detects a face of a person existing in the image captured by the image capturing unit. In the detected face of the person, the face of the person facing the direction of the robot is discriminated by the discriminating section, and the face that is closest to the robot among the discriminated faces is extracted by the extracting section. After extraction , The direction of imaging of the imaging unit, by changing in accordance with the position of the extracted face is characterized by tracking to continue imaging the extracted face.

  Such a speech recognition robot, when there are a plurality of persons in the imaging unit facing the direction in which the sound is generated, faces the direction of the robot and is located at the position closest to the robot. The speaker is regarded as a conversation target speaker and is controlled so as to keep facing the speaker. For this reason, it is possible to continue to face the speaker when performing a response with the speaker.

  Further, such a voice recognition robot may be configured to stop tracking when the extracted face goes out of an area defined around the robot. That is, when the person is away from the robot by a predetermined distance or more, it is determined that the person who is the conversation target is no longer the partner, and the tracking is stopped. When tracking is stopped in this way, if the voice from another speaker is received, the tracking target is extracted again so that the person speaking is detected as the next conversation target. It may be controlled.

  In addition, the sound source specifying unit that specifies the direction uttered by the speaker may be composed of one or more microphones having directivity. The voice recognition robot configured as described above can easily and accurately detect the direction in which the voice is uttered.

  Furthermore, the face detection unit specifies the positions of eyes and mouth included in the face of a person existing in the image captured by the imaging unit, and estimates the face contour based on these positions. Thus, the entire face may be detected. In this way, even if there is another person or object near the face of the person (speaker) and the outline of the person's face cannot be specified by simply measuring the distance from the robot, It is possible to specify the contour of the face only by detecting the positions of the eyes and mouth.

  Further, the positions of both eyes included in the face included in the image captured by the image capturing unit may be relatively determined from the viewpoint of the robot, and the direction in which the face is facing may be determined from the positional relationship. Good. In this way, the orientation of the speaker's face can be easily determined without considering the position of the entire face or the entire body of the speaker.

  Further, when the direction of the imaging unit to be imaged is changed so that the detected face is positioned substantially at the center in the captured image, the imaging unit is provided at a position corresponding to the face of the robot. In this case, it is possible to always speak toward the person (speaker) to be conversed. Therefore, it is possible to make it appear as if the conversation is being continued while looking at the face of the person to be conversed.

  Note that such a voice recognition robot may be configured to be movable within a predetermined area by providing a moving means. Since such a speech recognition robot can move by changing its own position in a state where the speaker is specified, it can be used for a robot that performs a guidance operation, for example.

  The present invention also provides a robot control method for controlling a voice recognition robot. Specifically, the present invention provides a step of specifying the direction in which the voice is generated, and receiving the spoken voice and recognizing the content. Determining the direction in which the sound is generated, capturing the captured image as image data, detecting the face of a person existing in the captured image, and determining the orientation of the detected face Determining a face facing the direction of the robot, extracting a face located at the closest distance from the robot among the determined faces, and imaging direction so as to track the extracted face A method for controlling a voice recognition robot, comprising: a step of controlling; and a step of tracking so as to continue imaging the extracted face It is intended to provide.

  By using such a voice recognition robot control method, even if there are a plurality of persons in the imaging unit facing the direction in which the voice is generated, the robot faces the robot and It is possible to control the robot so that the person located at the closest position is regarded as the speaker who is the subject of conversation and the speaker is kept facing. For this reason, it is possible to continue to face the speaker when performing a response with the speaker.

  In such a robot control method, tracking may be stopped when the extracted face is out of a predetermined distance from the robot. By controlling in this way, it is possible to detect that the person who was initially talking to the robot has stopped the conversation, and to search for the person who is the next conversation target and continue the conversation.

  As described above, according to the present invention, the voice recognition robot can be controlled to respond to the speaker in a state where the voice recognition robot is always directed to the speaker.

Embodiment 1 of the Invention
Hereinafter, a voice recognition robot according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 6.

  FIG. 1 shows a state where a plurality of people are moving in the room R and the voice recognition robot 1 is placed in the room R. As shown in FIG. 2, the voice recognition robot 1 shown in FIG. 1 includes a torso 10 fixed to the ground, a head 11, a right arm 12, and a left arm 13 connected to the torso 10. It is a humanoid robot constructed in the same way as the upper body. Hereinafter, each component will be described in detail.

  The body 10 is installed on the floor surface of the room R, and includes a control unit 100 that controls the operation of the voice recognition robot 1 and other functions. The control unit 100 is a control computer that includes an arithmetic processing unit, a memory, and the like that recognizes the contents of an input signal input from a microphone, which will be described later, and outputs the response data by voice after selecting appropriate response data. . The detailed configuration of the control unit 100 will be described later.

  The head 11 utters words to the outside, imaging units 111 and 112 for imaging a predetermined range in front of the voice recognition robot 1, sound source identification units 113 and 114 for listening to surrounding sounds, and And a speaker 115 for The imaging units 111 and 112 are optical cameras that each acquire optical information within a predetermined range, capture it as imaging data, and output it to the calculation unit 101, and are provided on the front left and right sides of the head 11.

  The sound source identification units 113 and 114 each have a plurality of so-called directional microphones each capable of acquiring sound from a certain direction, and a plurality of so-called directivity microphones are arranged in the horizontal direction. It is possible to roughly specify from which direction the signal is transmitted. Note that the sound source identification units 113 and 114 are provided on the left and right side surfaces of the head 11, acquire voices emitted around the voice recognition robot 1, acquire the voice data, and output the voice data to the arithmetic processing unit 101. .

  The speaker 115 outputs response data created by the control unit 100 to the outside in a predetermined direction and size, and is provided below the front surface of the head 11.

  The head 11 is connected to the body 10 so as to be rotatable in the left-right direction in a plane horizontal to the floor surface. To change the image of the surrounding environment.

  The right arm 12 and the left arm 13 are controlled by the arithmetic processing unit 101 included in the control unit 100 to control the amount of driving of the joints included in each arm according to a predetermined drive control program, and the joint driving angle of each joint is determined. Thus, a desired position and posture are taken.

  As shown in FIG. 3, the control unit 100 exists in an image captured by the speech recognition unit 101 that recognizes the content from the input speech data, and the imaging units 111 and 112 provided in the head 11. A face detection unit 102 for detecting the face of the person to be detected, a discrimination unit 103 for discriminating the position and orientation of the face detected by the face detection unit 102, an extraction unit 104 for extracting the detected face, and response data for outputting sound And a storage area (not shown) for storing predetermined programs and data.

  The voice recognition unit 101 converts the voice acquired from the sound source identification units 113 and 114 into voice data such as a WAVE file, divides the voice data into syllables, and stores a word database in which each syllable is stored in a storage area. Use to replace with a word. Then, the words included in the voice data and the word order thereof are analyzed, and the sentence closest to the analyzed voice data is selected from the many sentences stored in the storage area. If the degree of approximation between the selected text and the voice data is greater than or equal to a predetermined value, the analyzed voice data is recognized as the same content as the selected text, and indicates that the acquired voice is equal to the selected text Output a signal. If the closest sentence is less than the predetermined degree of approximation, it is determined that the corresponding sentence is not stored in the storage area, and a signal indicating that the content of the acquired voice has not been recognized is output.

  The face detection unit 102 detects only the face portion of a person from the image data obtained by imaging with the imaging units 111 and 112. The face detection unit 102 first extracts eyes and mouths (lips) included in a person's face, and estimates an edge corresponding to the contour of the face based on these positions. Then, an area surrounded by the estimated face contour is detected as a human face.

  It should be noted that, based on the position of the eyes included in each human face detected by the face detection unit 102, that is, the relative distance and relative direction from the robot, the detected face faces in the direction viewed from the robot. It can be estimated whether or not. Specifically, as shown in FIG. 4, the center positions E10 and E11 of the eyes (right eye E1 and left eye E2) included in the face of the person are specified, and the midpoint M on the line segment connecting the center positions is determined. Identify. Then, a direction P that includes the line segment connecting the center positions and is perpendicular to the line segment connecting the center positions from the midpoint M in a plane parallel to the floor surface is obtained. The line-of-sight direction of the left eye E2), that is, the direction in which the face including these eyes is facing. Then, the detected direction of each face is obtained, and a signal for each direction is output together with the relative position from the robot where each face exists.

  The discriminating unit 103 discriminates which face is directed to the robot itself from the relative position and the facing direction of each face included in the captured image detected by the face detecting unit 102. To do. Specifically, as shown in FIG. 5, the center position of each face determined based on the position of the eyes (right eye and left eye) with reference to its own position R (for example, the center point of the head 11), It is determined whether or not it includes its own position by combining with the direction the face is facing. Here, the direction in which each face is facing has a predetermined width, and in detail, the width is given by a minute angle (for example, 5 degrees) in the direction horizontal to the floor with each direction as the center. It is said. In this way, it is determined whether or not each face is facing the robot itself, and those not facing the robot itself are excluded from the extraction targets described later.

  The extraction unit 104 selects the face in the image that is determined to be facing the robot by the determination unit 103 and has the closest distance from the robot, and the contour portion and the inside thereof are included in the image data. Extract in Specifically, the center position of each face is specified, the distance between each center position and the central portion of the front surface of the robot head 11 is calculated, the face with the smallest calculated distance is selected, and the face The image data of the area occupied by is extracted. And even if the place where the image is taken changes as the head 11 rotates, the region that approximates the extracted image data is always specified. As a result, as long as the selected face exists in the image captured by the imaging unit, tracking can be performed. The distance between the center position of the face and the robot is calculated by, for example, a triangulation method.

  The speech synthesizer 105 reads the most appropriate response sentence data from among a large number of response sentence data groups stored in advance in the storage area corresponding to the acquired speech content recognized by the speech recognition unit 101. It converts into an audio file and outputs it outside through the speaker 115. Moreover, you may make it perform operation | movement, such as gesture, using an arm part (the right arm 12 and the left arm 13) with audio | voice output as needed.

  The speech recognition robot 1 configured as described above captures a person located near the front surface, identifies a person to whom he / she should respond to a plurality of persons included in the captured screen, and It recognizes the contents of the voice uttered by a person and outputs the voice according to the contents. Below, the procedure in which the voice recognition robot 1 specifies a person to whom it should respond based on the acquired voice will be described in detail with reference to the flowchart shown in FIG.

  First, when power is supplied, the voice recognition robot 1 prepares a state for acquiring voice from the surroundings (step 101). In this state, when a person existing around the voice recognition robot 1 speaks to the voice recognition robot 1, the voice recognition robot 1 acquires the voice uttered by the person using the sound source identification units 113 and 114 (step 102). At this time, the sound source identification units 113 and 114 identify the direction in which the acquired voice is uttered (relative direction as viewed from the robot). Then, the head 11 is rotated so that the front surface of the head 11 is positioned in the specified direction, and imaging is started (step 103).

  Image data obtained by imaging by the imaging units 111 and 112 is input to the control unit 100, and it is determined whether or not the face detection unit 102 can detect a human face in the image data (step 104). ). Here, if even one person's face can be detected in the image data, the direction of the detected face is determined by the determination unit 103, and it is determined whether or not there is a face facing the voice recognition robot 1. (Step 105). If there are faces facing the voice recognition robot 1, the distances of those faces from the voice recognition robot 1 are obtained, and the face existing at the closest position is selected (step 106). Conversely, if there is no face facing the voice recognition robot 1, it is determined that the voice recognition robot 1 is not in a talked state, and in order to prepare for acquiring voice again, step 101 is performed. Return.

  Next, when the face facing the voice recognition robot 1 that is present at the closest position is determined, the area occupied by the determined face in the image data is extracted, and the selected face is tracked. (Step 107). At this time, by rotating the head 11 so that the region extracted as the selected face is positioned at the approximate center of the image data, the voice recognition robot 1 is always directed toward the face of the person who speaks to the voice recognition robot 1. The head 11 is operated so that the front surface (the front surface of the voice recognition robot 1) is facing. Furthermore, since the sound source specifying units 113 and 114 also rotate simultaneously with the head 11, the sound from the direction specified as the sound source is always acquired.

  The operation of tracking the face of the person who speaks to the voice recognition robot 1 is continued until the person moves to a position separated from the voice recognition robot 1 by a predetermined distance or more. That is, it is determined whether the relative distance between the center position of the selected face and the voice recognition robot 1 is equal to or greater than a predetermined distance (step 108), and the conversation is continued as long as it is within the predetermined distance. Conversely, if it is determined that the distance is greater than the predetermined distance, tracking is canceled (step 109), and it is determined whether or not the next sound is to be acquired (step 110). And when performing voice acquisition continuously, it returns to step 101 and returns to the preparation state for acquiring a voice. When the voice acquisition is not continued, the power supply is stopped and the operation is terminated.

  In this way, while tracking the person who speaks to the voice recognition robot 1, the voice recognition robot 1 recognizes the words uttered by the person, and utters a response sentence corresponding to the contents, and continues the conversation. In addition, the conversation is stopped when the tracking is released.

  If the selected face is not imaged by the imaging units 111 and 112 of the voice recognition robot 1, for example, an obstacle exists between the robot and the person, or the person who has spoken to the voice recognition robot 1 is behind the robot. Also when moving to a position where the imaging units 111 and 112 cannot capture images, the speech recognition robot 1 determines that the distance from the selected face is equal to or greater than a predetermined distance, and stops tracking. That is, the tracking is stopped when the voice recognition robot 1 becomes unable to determine the face being tracked. At this time, the determination that the face being tracked can no longer be determined is confirmed by the fact that the area representing the selected face does not exist in the image data captured by the imaging units 111 and 112 for a very short time (for example, several seconds). Is preferably performed. In this way, when an obstacle passes instantaneously between the voice recognition robot 1 and the person, the conversation between the voice recognition robot 1 and the person is continued without interruption.

  In this embodiment, it is preferable to set the distance between the selected face and the robot to stop tracking to the longest distance at which a human face can be detected from the image captured by the imaging unit. Such a distance by which the human face can be detected by the imaging unit may be determined based on the resolution of the imaging unit, the brightness of the surroundings, and the like.

  In the present embodiment, the speech recognition unit converts the acquired speech into speech data, divides the speech data into syllables, and replaces the divided syllables with words. Is not limited to this, and can be used in place of many speech recognition methods currently used. Furthermore, the method of selecting a response sentence that responds to the recognized voice content is not limited to the above embodiment, and other methods can be applied.

  In the above-described embodiment, the face detection unit that detects only the face portion of the person from the image data extracts the eyes and mouth (lips) included in the face of the person, and based on these positions, Although the edge corresponding to the contour of the face is estimated, the present invention is not limited to this. For example, the face of a person may be detected based on the contour of the person's face itself or another part inside the face.

  Further, the extraction unit in the above-described embodiment calculates the distance between the center position of the face existing in the image and the robot, and selects the face having the minimum calculated distance. It is not limited to this. For example, a face whose face direction is the closest to the direction of each face with respect to the robot may be extracted. In this case, a straight line connecting the center of the face existing in the image and the robot is virtually drawn for each face, and an inclination angle formed by each straight line and the direction of the face is obtained. A method such as extracting the smallest face can be considered.

  In addition, the extraction unit may extract only the face portion existing in the image and perform tracking, and may set the entire body of the person including the face as a tracking target. In this way, even if the task of identifying the area occupied by the face portion fails, by extracting the body part, the location of the area occupied by the face portion is estimated and returned to the tracked state again. It becomes easy.

Embodiment 2 of the Invention
Next, with reference to FIGS. 7 and 8, an example of a voice recognition robot according to the second embodiment of the present invention, which is applied as a guide robot for guiding the inside of the room R, will be described. In the present embodiment, the same or similar components as those of the above-described embodiment are denoted by the same reference numerals, and the description thereof is omitted.

  As shown in FIG. 7, the speech recognition robot 1 ′ according to this embodiment includes a pair of wheels 21 and 22 and a caster 23 as moving means on a body 10 including a head 11 and an arm. In addition to being able to move according to instructions from the outside, it moves autonomously based on the external environment. The wheels 21 and 22 rotate independently to enable the robot 1 ′ to move left and right and turn. The wheels 21 and 22 and the caster 23 allow the robot 1 ′ to horizontally move the entire structure. The inside of the room R can be moved in a state where it is supported.

  Although not shown in the figure, a motor as a drive unit for driving the wheels 21 and 22 is independently provided in the body 10, and as a control unit for controlling these motors. CPU, a storage unit for storing map information about the room R, and the like are also provided. Based on the number of rotations of the wheels 21 and 22, the distance, speed, and direction of movement of the robot 1 ′ are calculated, the position information of the robot 1 ′ is acquired, and the self position is obtained together with the map information described above. ing.

  Procedure for the voice recognition robot 1 ′ configured in this way to image a person located in the vicinity of the front face and identify a person to whom he / she should respond to a plurality of persons included in the captured image This will be described in detail with reference to the flowchart shown in FIG. In the present embodiment, the procedure for detecting the face of a person existing in the captured image, the determination of the position and orientation of the detected face, the procedure for extracting the detected face, etc. are described in the above-described embodiments. It is assumed that the same method is used.

  Similar to the above-described embodiment, the voice recognition robot 1 ′ is installed at a predetermined initial position in the room R, and when power is supplied, prepares a state for acquiring voice from the surroundings (step 201). ). Note that the voice recognition robot 1 ′ stores map information in the room R in advance and recognizes the initial position.

  In this state, when a person existing around the voice recognition robot 1 ′ speaks to the voice recognition robot 1, the voice recognition robot 1 acquires the voice uttered by the person using the sound source identification units 113 and 114 (step 202). . At this time, the sound source identification units 113 and 114 identify the direction in which the acquired voice is uttered (relative direction as viewed from the robot), and drive the wheels 21 and 22 so that the front surface of the body 10 is positioned in the identified direction. Then, imaging is started using the imaging units 111 and 112 provided in the head 11 (step 203).

  Image data obtained by imaging by the imaging units 111 and 112 is input to the control unit 100, and it is determined whether or not the face detection unit 102 can detect a human face in the image data (step 204). ). Here, if even one person's face can be detected in the image data, the direction of the detected face is discriminated by the discriminating unit 103, and whether or not there is a face facing the voice recognition robot 1 'exists. Judgment is made (step 205). If there are faces facing the voice recognition robot 1 ', the distances between the faces from the voice recognition robot 1' are obtained, and the face existing at the closest position is selected (step 206). Conversely, if there is no face facing the voice recognition robot 11 ′, it is determined that the voice recognition robot 1 ′ is not in a state of talking to the voice recognition robot 11 ′, and the process returns to preparation for acquiring the voice again.

  Next, when a face that faces the voice recognition robot 1 ′ that is present at the closest position is determined, an area occupied by the determined face in the image data is extracted, and the selected face is tracked. (Step 207). At this time, by combining the rotation of the head 11 and the driving of the wheels 21 and 22 so that the region extracted as the selected face is located at the approximate center of the image data, the voice recognition robot 1 ′ is always used. The operation is performed so that the front surface of the head 11 of the speech recognition robot 1 ′ (the front surface of the speech recognition robot 1 ′) faces the direction of the face of the person talking. At the same time, since the sound source specifying units 113 and 114 also rotate simultaneously with the head 11, the sound from the direction specified as the sound source is always acquired.

  Then, the voice acquired from the tracked person is analyzed, and the target place to be guided is specified (step 208). When the place to be guided is specified, the voice recognition robot 1 keeps tracking the face of the person who talks to the voice recognition robot 1 ′, and the voice recognition robot 1 turns the front of the head 11 toward this person and starts from the tracked person. The wheels 21 and 22 are driven so as to be the distance, and the self-position is adjusted. Then, in order to guide to the place to guide in the room R, the wheels 21 and 22 are driven in a state where the distance from the tracked person is kept constant, and movement toward a predetermined place is started (step 209). ).

  During the guidance operation, it is detected whether there is a voice instruction to stop the guidance operation from the tracked person (step 210). If there is an instruction to stop the guiding operation, the guiding operation is stopped and the initial position in the room R is returned (step 310). If there is no instruction to stop the guiding operation, the predetermined guidance location is reached. Continue moving.

  Then, when reaching a predetermined place for guidance, it is confirmed that the guidance operation may be terminated for the tracked person (step 211), and if it can be terminated, the tracking is canceled and the initial position is returned. (Step 212). If the guidance operation is to be continued, the process returns to step 208, the voice from the tracked person is recognized, the location to be guided is acquired again, and the guidance operation is resumed. As described above, the voice recognition robot 1 ′ continues the guidance operation in the room R until an operation stop command is issued by the next voice acquisition determination (step 213).

  According to this embodiment, the robot can continue to move to a desired location in a state where the tracked person is always positioned at the center in the image and the position of the person to be guided is always confirmed. Therefore, the guidance operation can be performed in a state in which the movement of the robot is combined with the movement of the human being guided. Therefore, it is less likely that a human loses sight of the robot during the guidance operation, or conversely, the robot loses sight of the person to be tracked, and the guidance operation can be performed more reliably.

  In the above-described embodiment, the self-position is specified by obtaining the distance moved by the number of rotations of the wheel. Instead, the self-position of the robot is determined by using an external position recognition system. Then, the robot may be made to recognize its own position by transmitting a position signal to the robot. In this way, the robot can recognize a more accurate self-position in real time.

  The robot moving means is not limited to a combination of wheels and casters, and may be an inverted pendulum type moving means composed of only wheels, and is a walking type that moves by driving a leg. It may be a moving means.

  Also, as a means for recognizing the surrounding external environment provided in the robot, not only the imaging unit but also an optical camera such as a laser range sensor or CCD, or such a base station provided outside the robot is used. Environmental information may be transmitted to the robot.

BRIEF DESCRIPTION OF THE DRAWINGS It is the whole outline figure which shows a mode that the speech recognition robot which is 1st Embodiment based on this invention is provided indoors. FIG. 2 is a schematic view schematically showing the voice recognition robot shown in FIG. 1. FIG. 2 is a block diagram conceptually showing the inside of a control unit provided in the voice recognition robot shown in FIG. 1. It is a figure which shows a mode that the direction of each person's face is calculated | required by the face detection part with which the speech recognition robot shown in FIG. 1 was equipped. It is a figure which shows a mode that the discrimination | determination part with which the voice recognition robot shown in FIG. 1 was equipped discriminate | determines the face which has faced the direction of the robot. It is a flowchart which shows the procedure in which the speech recognition robot shown in FIG. 1 specifies the person who should respond based on the acquired voice. It is the schematic which shows roughly the speech recognition robot which is 2nd Embodiment which concerns on this invention. It is a flowchart which shows the procedure which the voice recognition robot shown in FIG. 7 performs guidance operation | movement, specifying the person who should respond based on the acquired audio | voice.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 1 '... Voice recognition robot 100 ... Control part 101 ... Voice recognition part 102 ... Face detection part 103 ... Discrimination part 104 ... Extraction part 105 ... Speech synthesis part 111 112, imaging unit 113, 114 ... sound source identification unit 115 ... speaker

Claims (9)

A sound source identifying unit for identifying the direction in which the sound is generated;
A voice recognition unit that receives the generated voice and recognizes its contents;
An imaging unit that captures images in at least two directions and acquires the captured images as image data;
A face detection unit for detecting the face of a person present in the captured image;
A discriminator for discriminating the position and orientation of the detected face;
An extraction unit for extracting the detected face;
A voice recognition robot comprising:
While capturing the direction in which the sound specified by the sound source specifying unit is generated,
A face of a person present in the image captured by the imaging unit is detected by the face detection unit;
Among the detected human faces, the discrimination unit determines the face of the person facing the robot,
Among the identified faces, the extraction unit extracts the face that is closest to the robot,
A voice recognition robot characterized in that the imaging direction of the imaging unit is changed in accordance with the position of the extracted face, and tracking is performed so that the extracted face is continuously imaged.   The voice recognition robot according to claim 1, wherein the tracking is stopped when the extracted face goes out of a region defined around the robot.   The voice recognition robot according to claim 1, wherein the sound source identification unit is composed of one or more microphones having directivity.   2. The face detection unit detects an entire face by specifying positions of eyes and mouth included in a human face and estimating a face outline based on these positions. 4. The voice recognition robot according to any one of 3.   5. The voice recognition robot according to claim 1, wherein the discrimination unit discriminates a direction of the detected face based on positions of both eyes included in the detected face from the robot.   The voice recognition robot according to claim 1, wherein the direction in which the detected face is captured is changed so that the detected face continues to be positioned at substantially the center of the captured image.   The voice recognition robot according to any one of claims 1 to 6, wherein the voice recognition robot further includes a moving unit, and is configured to be movable within a predetermined area. Identifying the direction in which the sound occurred;
Receiving the spoken voice and recognizing its contents;
Capturing in the direction in which the sound is generated, and acquiring the captured image as image data;
Detecting a human face present in the captured image;
Determining the direction of the detected face and determining the face facing the direction of the robot;
Extracting the face located at the closest distance from the robot among the determined faces;
Controlling the direction of imaging so as to track the extracted face;
Tracking so as to continue to capture the extracted face;
A method for controlling a speech recognition robot, comprising:   The method of controlling a voice recognition robot according to claim 8, further comprising a step of stopping tracking when the extracted face is out of a predetermined distance from the robot.