JP2007155985A - Robot and voice recognition device, and method for the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a robot and a voice recognition device, and a method for the same, capable of reducing malfunction generation by improving accuracy of voice recognition. <P>SOLUTION: The robot comprises: a microphone 14; a voice recognition section 50 for recognizing voice input from the microphone 14; a person detecting section 51a for detecting a person around a robot main body; a voice recognition effectiveness determination section 52 for determining effectiveness of a voice recognition result by the voice recognition section 50, when the person is detected by the person detecting section 51a; and a response action execution section 53 for executing response action corresponding to the voice recognition result, when the voice recognition result is determined to be effective by the voice recognition determination section 52. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a robot having a voice recognition function.

Conventional interactive robots for general homes are required to always respond to user calls in an environment where noise exists. However, it is very difficult to judge whether it is a user's command or a living noise such as TV / radio, and it is often the case that a living noise such as TV / radio is mistakenly recognized as a user's command. .
For such a problem, for example, Patent Document 1 proposes that speech recognition is always performed using a proper speech section as a solution to the above-described erroneous recognition. Specifically, the “speech recognition environment” such as a threshold used in voice segment detection, which is a process of extracting a voice segment to be recognized from the input of the microphone, is set as the type of voice recognition process being executed or the microphone. By switching according to the situation or the like, appropriate voice segment detection is always detected.
JP 2005-195834 A

  However, in the invention disclosed in Patent Document 1, since the voice recognition environment is switched only by the internal state of the robot (such as the voice recognition task being performed and the microphone selection status), for example, the robot is in the same internal state. In some cases, it was impossible to distinguish between voices emitted from radio etc. and voices generated by humans as commands, and there was still a problem that noise was mistakenly recognized as user commands and malfunctioned. .

  The present invention has been made to solve the above problems, and an object thereof is to provide a robot, a speech recognition apparatus, and a method thereof that can reduce the occurrence of malfunctions by improving the accuracy of speech recognition. To do.

In order to solve the above problems, the present invention employs the following means.
The present invention provides a voice input means, a voice recognition means for recognizing voice input from the voice input means, a person detection means for detecting a person around the robot body, and a person detected by the person detection means A determination unit that validates the speech recognition result by the speech recognition unit, and a response operation execution unit that executes a response operation corresponding to the speech recognition result when the speech recognition result is validated by the determination unit. A robot comprising:

According to such a configuration, the person detection unit detects whether or not a person around the robot body exists, and the detection result is supplied to the determination unit. When the person is detected by the person detection unit, the determination unit validates the voice recognition result by the voice recognition unit and outputs the voice recognition result to the response operation execution unit. Thereby, the response operation corresponding to the voice recognition result is executed by the response operation executing means.
In this case, whether or not the voice recognition result is valid is determined based on whether or not there is a person in the vicinity, so that the probability of erroneously recognizing life noise such as television and radio as voice input from the user is reduced. Can do. Furthermore, since a conversation or the like is performed based on the highly accurate speech recognition result, the probability of malfunction can be greatly reduced.

  In the robot described above, the response operation executing unit may execute an alternative response operation indicating that a sound is detected when the speech recognition result is invalidated by the determination unit.

In this way, when the speech recognition result is invalidated by the determination means, an alternative response operation indicating that a sound has been detected is executed. It can be naturally melted into the space.
The alternative operation is preferably an operation that can be completed in a short time (for example, several seconds or several tens of seconds), such as performing an operation that completes with one movement that emits a voice composed of several characters. By performing such a simple operation, for example, even when a voice input from the user is performed immediately after that, it is possible to promptly respond to the voice input of the user.

  In the robot described above, the determination unit may validate the voice recognition result regardless of the detection result of the person detection unit when an important term related to a command is received as the voice recognition result.

  As described above, when an important term related to an instruction, for example, a keyword for executing various response operations is input, the determination unit determines that the voice recognition result is valid regardless of the detection result of the person detection unit. Therefore, it becomes possible to accept a user request with a high probability.

  In the robot described above, the response operation executing unit performs the response operation being executed when a voice recognition result valid by the determination unit is not output over a predetermined period during the execution period of the response operation. It is good also as ending.

  According to such a configuration, when the voice recognition result is not output to the response operation executing unit over a predetermined period, by determining that the voice input from the user has not been performed over the predetermined period, The response operation being executed by the response operation executing means is terminated. As a result, even if the user stops speaking during the execution of the response operation, for example, in the middle of the conversation, and moves away from the periphery of the robot body, the response operation is performed after a predetermined period from the latest voice input. It can be terminated.

  The robot may further include a face detection unit that detects the face of a person around the robot body, and the determination unit performs voice recognition by the voice recognition unit when the face of the person is detected. It is good also as validating a result.

  According to such a configuration, the determination unit determines whether or not the voice recognition result by the voice recognition unit is valid based on whether or not a person's face is detected. It is possible to respond to the voice input from the user only when facing the direction. As a result, life noise is not erroneously recognized as voice input from the user, and malfunctions can be reduced.

  In the robot described above, the determination unit may be configured so that an elapsed time from when the response operation is performed by the response operation execution unit or after the execution of the response operation is within an estimation period for estimating that a person is in the vicinity. In some cases, the voice recognition result may be validated regardless of the detection result of the face detection means.

  As described above, when the estimation period has not elapsed since the most recent determination that the input is a voice input from the user, and the voice recognition result is input to the determination unit again from the voice recognition unit, this voice recognition is performed. Since it is determined that the information is from the user, it is possible to determine whether the speech recognition result is valid or invalid by an easy determination process.

  In the robot described above, the person detection unit may include an imaging unit that captures the periphery of the robot body, and may detect a person from image information acquired by the imaging unit. An omnidirectional camera provided on the top of the robot body may be employed.

  In this way, by adopting an omnidirectional camera provided at the top of the robot body as an imaging means for acquiring surrounding information as an image, the surrounding 360 is always independent of the orientation of the face of the robot. It becomes possible to obtain information of °. As a result, stable person detection can be realized.

In the robot described above, the person detection unit may include a contact detection unit that detects contact, or may include a distance sensor that measures a distance to an object.
By using information obtained by such a contact detection means, a distance sensor, or the like, it is possible to easily detect a person.

  In the robot described above, the person detection means includes storage means for storing information from the imaging device, the contact detection means, and the distance sensor, and the information in the past predetermined period stored in the storage means. Based on this, a person may be detected.

  As described above, by detecting a person based on information in a past predetermined period (for example, about 10 seconds to about a dozen seconds or more), it becomes possible to grasp the surrounding situation in more detail, so that the accuracy of person detection is improved. It becomes possible to improve.

  The present invention provides a voice input means, a voice recognition means for recognizing voice input from the voice input means, a person detection means for detecting a person around the robot body, and a person detected by the person detection means In addition, when the speech recognition result is validated by the determination unit, the parameter changing unit that lowers the threshold of the certainty of correct answer of the speech recognition unit, and the response operation execution that executes the response operation corresponding to the speech recognition result And a robot comprising the means.

  The present invention provides a voice acquisition process for acquiring an input voice, a voice recognition process for recognizing the input voice, a person detection process for detecting a person around the robot body, and a person detected in the vicinity of the robot body. There is provided a speech recognition method comprising a determination step of validating a speech recognition result in the speech recognition step.

  The present invention provides a voice acquisition process for acquiring an input voice, a voice recognition process for recognizing the input voice, a person detection process for detecting a person around the robot body, and a person detected in the vicinity of the robot body. There is provided a speech recognition method including a parameter changing process for reducing a threshold of a certainty of correct answer in a speech recognition process.

  According to the present invention, since it is possible to improve the accuracy of voice recognition, it is possible to reduce the occurrence of malfunction.

Hereinafter, an embodiment of a robot according to the present invention will be described with reference to the drawings.
FIG. 1 is a front view of a robot according to an embodiment of the present invention, and FIG. 2 is a left side view of the robot shown in FIG.
As shown in FIGS. 1 and 2, the robot body 1 includes a head 2, a chest 3 that supports the head 2 from below, a right arm 4 a provided on the right side of the chest 3, and a chest 3. A left arm portion 4b provided on the left side, a waist portion 5 connected below the chest portion 3, a skirt portion 6 connected below the waist portion 5, and a leg portion 7 connected below the skirt portion 6; Is provided.

One omnidirectional camera 11 is provided near the top of the head 2. A plurality of infrared LEDs 12 are arranged on the ring at predetermined intervals along the outer periphery of the omnidirectional camera 11.
In the vicinity of the center of the front surface of the head 2, as shown in FIG. 1, the front camera 13 for imaging the front is one on the right side when viewed from the front, and the microphone 14 is one on the left side when viewed from the front. Is provided.

  One monitor 15 is provided near the center of the front surface of the chest 3. One ultrasonic distance sensor 16 for detecting a person is provided above the monitor 15. One power switch 17 is provided below the monitor 15. Above the ultrasonic distance sensor 16, two speakers 18 are provided one on each side. In addition, as shown in FIG. 2, a backpack 33 that can store luggage is provided on the back of the chest 3. The school bag 33 is provided with an opening / closing door 33a that can be rotated around a hinge provided at the top. As shown in FIG. 1, one shoulder switch 19 that functions as a man-machine interface is provided on each of the left and right shoulders of the chest 3. For the shoulder switch 19, for example, a touch sensor is employed.

  A multi-joint structure is adopted for the right arm portion 4a and the left arm portion 4b. In the right arm portion 4a and the left arm portion 4b, side switches 20 are provided in the vicinity of the connection portion with the chest portion 3 to detect the pinching of a body or an object and stop the movement of the arm. As shown in FIG. 1, a handshake switch 21 that functions as a man-machine interface is built in the palm of the right arm 4a. For the side switch 20 and the handshake switch 21, for example, a pressure sensor is employed.

  In the vicinity of the center of the front surface of the waist 5, one ultrasonic distance sensor 22 for detecting a person is provided on each side. Below these ultrasonic distance sensors 22, a sensor region 24 in which a plurality of infrared sensors 23 are arranged is provided. These infrared sensors 22 are for detecting an obstacle or the like in the lower front of the robot body 1. As shown in FIG. 1 and FIG. 2, a total of four microphones 25 are provided below the waist 5 for detecting the sound source direction, one on the left and one on the front and back. As shown in FIG. 2, one handle portion 26 used for lifting the main body is provided on each of the left and right sides of the waist portion 5. The handle 26 is a recess so that the operator's hand can be inserted.

  Below the front surface of the skirt portion 6, a total of three infrared sensors 27 for detecting a step are provided in the center and on the left and right. As shown in FIG. 2, a charging connector 28 is provided on the back surface of the skirt portion 6.

As shown in FIG. 1, one infrared sensor 29 for detecting a lateral distance is provided on the front surface of the leg portion 7 on the left and right sides. These infrared sensors 29 are mainly used for level difference detection.
As shown in FIG. 2, a hook 30 for fixing the position of the robot body 1 to the charging station is provided on the back surface of the leg portion 7. The leg portion 7 is a carriage provided with traveling wheels 31 and four ball casters 32.

Such a robot has a configuration capable of moving independently in a work space by supplying power from a battery built in the robot body 1, and coexists with a human being as a work space indoors. For example, it is used to provide various services for assisting, supporting, and caring for the lives of users such as robot owners and operators in general households.
Therefore, in addition to the conversation function for realizing the conversation with the user, the robot 1 has a function of watching the user's action, assisting the user's action, and acting with the user. Such a function is realized by, for example, a control device including a microcomputer or the like built in the robot body 1 described later. Various cameras and various sensors shown in FIGS. 1 and 2 are connected to the control device, acquire image information from the cameras and sensor detection information from the sensors, and execute various programs based on these information. By executing this, the various functions described above are realized. The shape of the robot body 1 is not limited to the shape shown in FIGS. 1 and 2, and various shapes such as a model imitating an animal for pets can be adopted.

[Voice Recognition Apparatus According to First Embodiment]
Next, the speech recognition function that is a characteristic part of the present invention will be described. The voice recognition function is a function necessary for realizing the conversation function described above, and is realized by a voice recognition device provided in the control device described above.
As shown in FIG. 3, the speech recognition apparatus according to the present embodiment includes a speech recognition unit (speech recognition unit) 50 that recognizes speech input from the microphone 14 and a person detection unit that detects a person around the robot body ( (Person detection means) 51a, and when a person is detected by the person detection section 51a, a voice recognition acceptance / rejection judgment section (judgment means) 52 that validates the voice recognition result by the voice recognition section 50, and a voice recognition acceptance / rejection judgment section 52 And a response operation execution unit (response operation execution means) 53 that executes a response operation according to the determination result of.

The microphone 14 captures user's voice and life noises such as telephone bells, doorbells, TV / radio, etc., converts these voices into electrical signals, and outputs them to the voice recognition unit 50.
The voice recognition unit 50 collates the electrical signal output from the microphone 14 with a plurality of dictionary data held in advance, and when a collation result exceeding a predetermined correct answer certainty factor is obtained, the collation result is voice-recognized. As a result, it is output to the speech recognition acceptance / rejection determination unit 52. Here, the predetermined correct answer certainty factor is a reference value for determining that the voice data from the microphone 14 is the same as the collated dictionary data.

The person detection unit 51 a includes a camera 61, a contact sensor 62, a human detection sensor 63, a processing unit 64, and a data holding unit (storage unit) 65.
The camera 61 is, for example, the omnidirectional camera 11 or the front camera 13 shown in FIGS. 1 and 2, and acquires information about 360 ° around the robot body 1 and surrounding information about the front of the head 2 as an image. The image information is output to the processing unit 64. In particular, by adopting the omnidirectional camera 11 as a camera for acquiring surrounding image information, it is possible to always capture stable surrounding information regardless of the front direction of the robot head 2. Thereby, the accuracy of person detection can be improved.

  The contact sensor 62 is provided, for example, on the shoulder switch 19 provided on the chest 3 of the robot body 1 shown in FIGS. 1 and 2, the handshake switch 21 provided on the right arm 4a, the side switch 20, and the left arm 4b. The side switch 20 and the like, and these sensors output a contact detection signal to the processing unit 64 when operation information from the user is obtained.

The human detection sensor 63 includes, for example, the ultrasonic distance sensors 16 and 22 shown in FIGS. 1 and 2, the microphone 25 for detecting the direction of the sound source, and the like. Output to 64.
The processing unit 64 creates inter-frame difference data by processing the captured image acquired from the camera 61, detects whether there is something that moves based on these difference data, and if there is something that moves, It is determined that there is a person around the robot body. Alternatively, the processing unit 64 may perform person detection by extracting a person's face, body silhouette, and the like from the acquired image information.

  Further, when the processing unit 64 receives a contact detection signal from the contact sensor 62, the processing unit 64 determines that there is a person around the robot body. Further, the processing unit 64 detects the distance to the obstacle ahead of the robot body 1 from the sensor signals from the ultrasonic distance sensors 16 and 22 constituting the human detection sensor 63, and the distance is within a predetermined range. If it is determined that there is a person around the robot body 1. Note that audio information from the microphone is used to detect the direction in which the person is present.

  When the processing unit 64 performs the above-described various determinations based on the information acquired from the camera 61, the contact sensor 62, and the human detection sensor 63, when it is determined that there is a person around the robot body in any of the determination results. Outputs the person detection information indicating that there is a person in the surroundings to the voice recognition acceptance / rejection determination unit 52, and if it is determined that there is no person around the robot body in any of the determination results, there is no person in the vicinity. Is output to the voice recognition acceptance / rejection determination unit 52.

  The data storage unit 65 predetermines the data used by the processing unit 64 for human detection, in other words, the information acquired by the processing unit 64 from the camera 61, the contact sensor 62, and the human detection sensor 63 using a FIFO (First In First Out) method. It will be held for the period.

  When the speech recognition acceptance / rejection determination unit 52 receives a speech recognition result from the speech recognition unit 50, the speech recognition acceptance / rejection determination unit 52 determines whether or not person detection information indicating that there is a person around is input from the person detection unit 51a. As a result, when the person detection information indicating that there is a person around is input, it is determined that the voice recognition result by the voice recognition unit 50 is valid, that is, the voice based on the voice recognition result is input by the user. The voice recognition result is output to the response operation execution unit 53. On the other hand, the voice recognition acceptance / rejection determination unit 52 invalidates the voice recognition result by the voice recognition unit 50 when the person detection information indicating that there is no person is input from the person detection unit 51a, that is, a television or the like. And the determination result is output to the response operation execution unit 53.

  The response operation execution unit 53 includes a conversation scenario execution unit 66 and an alternative operation execution unit 67. The response operation executing unit 53 has, for example, a small microcomputer and an application program in which procedures for realizing various response operations such as a conversation operation and an alternative response operation are described. By reading and executing an application program corresponding to the information received from 52, various functions by the conversation scenario execution unit 66 and the alternative operation execution unit 67 as described later are realized.

The conversation scenario execution unit 66 creates a conversation scenario according to the voice recognition result received from the voice recognition acceptance / rejection determination unit 52, and outputs a voice signal based on the synthesized voice data to the speaker 18 included in the robot main body 1. Speaks according to the synthesized voice data.
The alternative action execution unit 67 realizes an alternative action indicating that surrounding sound has been detected when an invalidity determination is received from the voice recognition acceptance / rejection determination part 52.
As the alternative operation, for example, an operation or speech that is completed in a short time (several seconds to several tens of seconds) is preferable. For example, the movement can be completed by one movement such as raising the neck, full neck, looking around, opening and closing eyes, etc. For speech, "Fu", "Cuppi" An example is composed of several characters such as "N-".

In such a speech recognition device, when sound is received by the microphone 14 and an electrical signal based on this sound is output to the speech recognition unit 50, the speech recognition unit 50 performs speech recognition based on this electrical signal, The voice recognition result is output to the voice recognition acceptance / rejection determination unit 52.
On the other hand, in the person detection unit 51a, image information of the camera 61 and various sensor information from the contact sensor 62 and the human detection sensor 63 are periodically output to the processing unit 64, and the processing unit 64 detects the person based on these information. Are performed periodically (at predetermined time intervals). The determination result of the person detection by the processing unit 64 is output to the voice recognition acceptance / rejection determination unit 52 as person detection information.

  When receiving the voice recognition result from the voice recognition unit 50, the voice recognition acceptance / rejection determination unit 52 refers to the person detection information from the person detection unit 51a. As a result, if the person detection information is information indicating that there is a person in the surroundings, the voice recognition result is determined to be input from the user, the voice recognition result is validated, and the voice recognition result is responded. The data is output to the conversation scenario execution unit 66 of the execution unit 53. The conversation scenario execution unit 66 that has received the voice recognition result 52 outputs scenario voice data corresponding to the voice recognition result 52 to the speaker 18. As a result, an appropriate utterance corresponding to the voice input from the user is made.

  On the other hand, if the person detection information when the voice recognition result is received from the voice recognition unit 50 is information indicating that there is no person in the vicinity, the voice recognition acceptance / rejection determination unit 52 displays the voice recognition result as life. It is determined that the noise is due to noise, and the determination result is output to the alternative operation execution unit 67. Receiving this determination result, the substitute operation execution unit 67 performs a predetermined operation, for example, an operation such as a neck bend.

  As described above, according to the speech recognition apparatus according to the present embodiment, the result of speech recognition is compared with the surrounding situation to determine whether the input is from the user. The probability of erroneously recognizing life noise such as voice recognition by the user can be reduced. Then, since an operation such as a conversation is executed based on such a speech recognition result, the probability of malfunction can be reduced.

In the present embodiment, it is also possible to detect whether or not a person is around by taking into account past image information and sensor information.
Specifically, the processing unit 64 of the person detection unit 51a may perform person detection comprehensively using information in the past predetermined period stored in the data storage unit 65.
The voice recognition acceptance / rejection determination unit 52 may hold the person detection information from the person detection unit 51a for a predetermined period in the past, and may determine whether the voice recognition result is valid or invalid based on the information.
In this way, it is possible to improve the accuracy of person detection by determining whether or not there is a person based on information over the past predetermined period as well as when the voice is input.

[Second Embodiment]
Next, a speech recognition apparatus according to the second embodiment of the present invention will be described. As shown in FIG. 4, the speech recognition apparatus according to the present embodiment is different from the first embodiment in that the processing unit 64 of the person detection unit 51b has a face detection function (not shown) for detecting a person's face. The face detection result is output to the voice recognition acceptance / rejection determination unit 52 as face detection information. Hereinafter, with respect to the speech recognition apparatus of the present embodiment, description of points that are common to the first embodiment will be omitted, and only differences will be described.

In FIG. 4, the processing unit 64 of the person detection unit 51 b includes a face extraction function (not shown) that extracts face information from image information received from the camera 61. As a technique for extracting or recognizing a person's face from image information, a known technique can be employed. For example, face detection can be performed by extracting two ellipse parts from the image information acquired from the camera 61 and recognizing the eyes by examining their distance, size, color, and the like. Further, a technique for detecting a face by extracting a face outline may be employed.
The person detection unit 51b periodically performs the above-described person detection process and face detection process, and outputs the detection results to the voice identification acceptance / rejection determination unit 52 as person detection information and face detection information.

  The voice recognition acceptance / rejection determination unit 52 refers to the person detection information and the face detection information when the voice recognition result is received from the voice recognition unit 50, and when there is a person around and a face is detected, The recognition result is determined to be valid, and the speech recognition result is output to the conversation scenario execution unit 66. On the other hand, if there is a person around but no face is detected, or if neither a person nor a face is detected, the voice recognition result is determined to be invalid, and this determination result is used as an alternative action execution unit 67. Output to.

  As described above, according to the speech recognition apparatus according to the present embodiment, when a person is not only in the vicinity but also when the face of the person can be detected, the speech recognition result can be obtained by speech input from the user. Since it is determined that there is, it is possible to further reduce the probability of misrecognizing life noise as voice input from the user, and as a result, it is possible to further reduce the malfunction of the robot.

Note that the face detection described above may be performed by referring to information of a past predetermined period, as in the first embodiment described above.
Further, the speech recognition acceptance / rejection determination unit 52 relates to the speech recognition result input from the speech recognition unit 50 within the estimation period in which it can be estimated that the person is around from the latest determination result that the speech recognition result is valid. May automatically determine that the speech recognition result is valid. In other words, when the estimated period has not elapsed since the previous voice input from the user, it is clear that there is a user around. In this case, the voice from the voice recognition unit 50 is automatically set. By treating the recognition result as valid, it is possible to easily make a determination without performing complicated processing.

[Third Embodiment]
Next, a speech recognition apparatus according to the third embodiment of the present invention will be described. The speech recognition apparatus according to the present embodiment is suitable for use in an aspect in which, for example, the utterance by the conversation scenario execution unit 66 proceeds step by step in an interactive manner with the user.
Here, the case of proceeding step by step in an interactive format means that, for example, when “message” is input from the user, an utterance “What is the message? In response to this question, when the user further inputs a voice, for example, “Mr. A”, the conversation scenario execution unit 66 responds to “I want to tell Mr. A. Are you sure?” In addition, it means that the conversation progresses with the user. The above is an example, and the content of the conversation is not limited.

Then, as described above, when the conversation proceeds with the user, a standby period (hereinafter referred to as a response from the user corresponding to the utterance after the utterance is made by the conversation scenario executing unit 66). The state of the conversation scenario execution unit 66 during this period is referred to as “standby state”).
The voice recognition apparatus according to the present embodiment is characterized by the processing of the conversation scenario execution unit 66 in this standby state.

  Specifically, as shown in FIG. 5, in the speech recognition apparatus of the present embodiment, a configuration in which the person detection unit 51 c outputs person detection information to the speech recognition acceptance / rejection determination unit 52 and the response operation execution unit 53. Take. With this configuration, the conversation scenario execution unit 66 has not received a voice recognition result from the voice recognition acceptance / rejection determination unit 52 over a predetermined period after entering the standby state, and the person detection information When the information indicating that there is no person in the surroundings is continuously input, the standby state is canceled by transitioning to the initial state by terminating the currently executing response operation. That is, it returns to the original state where no voice input is made by the user.

As described above, according to the speech recognition apparatus according to the present embodiment, in the conversation scenario execution unit 66 in the standby state, from the speech recognition acceptance / rejection determination unit 52 for a predetermined period after entering the standby state. If the voice recognition result is not received and information indicating that there is no person in the surroundings is continuously input as the person detection information, the standby state is set by terminating the currently executed response action. Release and transition to the initial state.
Thereby, for example, when the user stops the conversation in the middle of the conversation and moves away from the periphery of the robot body 1, this can be detected promptly and the conversation scenario execution unit 66 can release the standby state. It becomes possible. As a result, for example, it is possible to avoid a state where the user has been waiting for a voice input for a long time.

[Fourth Embodiment]
Next, a speech recognition apparatus according to the fourth embodiment of the present invention will be described. The voice recognition apparatus according to the present embodiment is different from the first embodiment in that the voice recognition acceptance / rejection determination unit 52 shown in FIG. 1 is not provided as shown in FIG. The person detection information is directly output to the unit 50, and the voice recognition result is directly output from the voice recognition unit 50 to the conversation scenario execution unit 66.
Hereinafter, with respect to the speech recognition apparatus of the present embodiment, description of points that are common to the first embodiment will be omitted, and only differences will be described.

  In FIG. 6, the processing unit 64 of the person detection unit 51 d performs person detection based on information input from the camera 61, the contact sensor 62, the person detection sensor 63, and the like, and detects whether there is a person around. The information is output to the voice recognition unit 50 ′.

The voice recognition unit 50 ′ includes a parameter changing unit (not shown) that varies a threshold of accuracy certainty used when performing voice recognition based on the person detection information from the person detection unit 51d. When it is determined by 51d that the person is in the vicinity, the threshold value of the correct answer reliability is set low. On the other hand, when it is determined that the person is not in the vicinity, the threshold value of the correct answer reliability is set high. Thereby, the reliability of speech recognition when it is determined that the person is not in the vicinity is set more strictly than when the person is determined as being around.
The voice recognition unit 50 ′ performs voice recognition based on the correct answer certainty set by the built-in parameter changing unit. That is, when the collation result between the voice information input from the microphone 14 and the dictionary data held in advance is equal to or higher than the correct answer certainty set by the parameter changing unit, the voice information is the voice of the dictionary data. It is determined that they are the same, and the dictionary data information at this time is output to the conversation scenario execution unit 66 as a voice recognition result.

The conversation scenario execution unit 66 outputs scenario voice data corresponding to the voice recognition result received from the voice recognition unit 50 ′ to the speaker 18. As a result, an appropriate utterance corresponding to the voice input from the user is made.
On the other hand, when the speech recognition unit 50 ′ matches the speech information input from the microphone 14 and the dictionary data held in advance with a degree of correctness certainty set by the parameter changing unit, Since the voice information fails to be recognized as a result of voice recognition, no information is output, so that the conversation scenario execution unit 66 does not operate.

  As described above, in the speech recognition apparatus according to the present embodiment, the correct confidence value used for speech recognition is changed depending on whether or not there is a person around the robot body. The success rate of speech recognition is increased by performing speech recognition based on the accuracy confidence set low, and the speech recognition success is achieved by performing speech recognition based on the correct confidence confidence set higher when there is no person. The rate is decreasing. As a result, it is possible to reduce the probability of erroneously recognizing life noise such as television as voice input from the user, and to prevent malfunction of the robot. Furthermore, according to the present embodiment, as shown in FIG. 1, since the voice recognition acceptance / rejection determination unit 52 is not required, the configuration can be simplified, and the apparatus can be reduced in size and weight. It becomes possible.

In addition, it is good also as combining the structure in each embodiment mentioned above. For example, the face detection information may be added to the information output from the person detection unit 51b to the response operation execution unit 53 by combining the second embodiment and the third embodiment.
In the fourth embodiment, face detection information can be added to information output from the processing unit 64 to the voice recognition unit 50 '.
In the speech recognition acceptance / rejection determination unit 52, if the speech recognizer result input from the speech recognition unit 50 is an important term related to a command to the robot, the speech recognition regardless of information from the person detection unit. It is good also as judging that a recognition result is effective.
Further, the determination criterion of the speech recognition acceptance / rejection determination unit 52 may be changed according to the importance of the request from the user. For example, if the voice content based on the voice recognition result of the voice recognition unit 50 is highly important, it is determined that the face or person is effective even if no face or person is detected. If the face is not visible, it may be determined to be invalid.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes design changes and the like within a scope not departing from the gist of the present invention.

It is a front view of the robot which concerns on one Embodiment of this invention. It is a left view of the robot shown in FIG. It is a figure which shows the structure of the speech recognition apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the structure of the speech recognition apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows the structure of the speech recognition apparatus which concerns on the 3rd Embodiment of this invention. It is a figure which shows the structure of the speech recognition apparatus which concerns on the 4th Embodiment of this invention.

Explanation of symbols

14 microphone 61 camera 62 contact sensor 63 human detection sensor 64 processing unit 65 data storage unit 50, 50 'voice recognition unit 51a to 51d person detection unit 52 voice recognition acceptance / rejection determination unit 53 response operation execution unit 66 conversation scenario execution unit 67 alternative operation Execution part

Claims (14)

Voice input means;
Voice recognition means for recognizing voice input from the voice input means;
A person detection means for detecting a person around the robot body;
A determination unit that validates a voice recognition result by the voice recognition unit when a person is detected by the person detection unit;
And a response action executing means for executing a response action corresponding to the voice recognition result when the voice recognition result is validated by the determination means.   The robot according to claim 1, wherein the response operation executing unit executes an alternative response operation indicating that a sound is detected when the voice recognition result is invalidated by the determination unit.   3. The determination unit according to claim 1, wherein the determination unit validates the voice recognition result regardless of a detection result of the person detection unit when an important term related to a command is received as the voice recognition result. The robot described.   The response operation executing means terminates the response operation being executed when a voice recognition result valid by the determination means is not output over a predetermined period during the execution period of the response operation. The robot according to any one of claims 1 to 3. Equipped with a face detection means for detecting the face of a person around the robot body,
The robot according to any one of claims 1 to 4, wherein the determination unit validates a voice recognition result by the voice recognition unit when the face of the person is detected.   In the case where the response operation is executed by the response operation execution unit or the elapsed time from the end of the execution of the response operation is within an estimation period in which it is estimated that the person is in the vicinity, The robot according to any one of claims 1 to 5, wherein the voice recognition result is valid regardless of a detection result of the face detection unit.   The robot according to any one of claims 1 to 6, wherein the person detecting means includes an imaging means for photographing the periphery of the robot body, and detects a person from image information acquired by the imaging means.   The robot according to claim 7, wherein the imaging device is an omnidirectional camera provided at the top of the robot body.   The robot according to any one of claims 1 to 8, wherein the person detection means includes contact detection means for detecting contact.   The robot according to claim 1, wherein the person detection unit includes a distance sensor that measures a distance to an object.   The person detection means includes storage means for storing information from the imaging device, the contact detection means, and the distance sensor, and a person is detected based on information in a past predetermined period stored in the storage means. The robot according to claim 10 to be detected. Voice input means;
Voice recognition means for recognizing voice input from the voice input means;
A person detection means for detecting a person around the robot body;
A parameter changing means for lowering a threshold of correctness confidence of the voice recognition means when a person is detected by the person detecting means;
And a response action executing means for executing a response action corresponding to the voice recognition result when the voice recognition result is validated by the determination means. An audio acquisition process for acquiring input audio;
A speech recognition process for recognizing the input speech;
A person detection process for detecting a person around the robot body,
A speech recognition method comprising: a determination process for validating a speech recognition result in the speech recognition process when a person is detected around the robot body. An audio acquisition process for acquiring input audio;
A speech recognition process for recognizing the input speech;
A person detection process for detecting a person around the robot body,
A parameter recognition process comprising: a parameter changing process for lowering a threshold of correct confidence in the voice recognition process when a person is detected around the robot body.

Images