JP2001215990A - Robot hearing device - Google Patents

Abstract

(57) [Problem] To provide an active auditory perception capable of collecting sound from an external target and performing active perception without being affected by noise generated inside the robot such as a drive mechanism. Provide equipment. SOLUTION: A soundproof exterior 14 covering at least a head 13 of the robot, a pair of external microphones 16 provided at both ear positions outside the exterior and mainly collecting external sounds, and mainly provided inside the exterior A process of canceling a noise signal from an internal noise source from an acoustic signal from an external microphone based on a pair of internal microphones 17 that collects noise from an internal noise source and an external microphone and a signal from the internal microphone Circuits 18 and 19, wherein the processing circuit comprises:
The robot hearing device 10 is configured to cancel the noise of the acoustic signal from the external microphone by arithmetically processing the acoustic signal from the external microphone and the acoustic signal from the internal microphone.
Is configured.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a hearing device for a robot, particularly a human or animal robot.

[0002]

2. Description of the Related Art In recent years, attention has been paid to active perception of sight and hearing in such humanoid or animal type robots. The active perception is to control a posture of a perception device that is responsible for perception such as robot vision and robot hearing by a drive mechanism, for example, a head that supports these perception devices so as to follow a target to be perceived.

Here, regarding active vision, at least a camera, which is a perceptual device, holds its optical axis toward a target by attitude control by a driving mechanism, and automatically focuses, zooms in, and zooms out on the target. The target is imaged by the camera by performing the above-mentioned operations, and various studies have been made.

On the other hand, regarding active hearing, at least a microphone, which is a perceptual device, maintains its directivity toward a target by attitude control by a driving mechanism, and sounds from the target are collected by the microphone. At this time, the disadvantage of active hearing is that the microphone picks up the operating sound of the drive mechanism while the drive mechanism is operating, so that relatively loud noise is mixed in the sound from the target, and Sound may not be recognized.

[0005]

However, in such a hearing study with the driving mechanism stopped, especially when the target is moving, the so-called active hearing while following the movement of the target is required. It was difficult to do.

Further, not only the above-mentioned drive mechanism but also various operation sounds generated inside the robot and noises generated constantly are collected by the microphone as a hearing device, so that a complete active hearing is similarly obtained. It was difficult.

In view of the above, the present invention enables active perception by collecting sound from an external target without being affected by noise generated inside the robot such as a drive mechanism. It is an object of the present invention to provide a robot hearing device.

[0008]

In order to achieve the above object, according to the present invention, a robot provided with a noise source inside has a soundproof exterior covering at least a part of the robot and an exterior provided outside the exterior. At least one external microphone that mainly collects external sounds, at least one internal microphone that is provided inside the exterior and mainly collects noise from an internal noise source, and the external microphone and the internal microphone. A processing circuit that cancels a noise signal from an internal noise source from an audio signal from an external microphone based on the signal of the external microphone, and the processing circuit converts the audio signal from the external microphone and the audio signal from the internal microphone. It is characterized in that arithmetic processing cancels noise of an acoustic signal from an external microphone.

According to another aspect of the present invention, there is provided a human-type or animal-type robot having a noise generating source such as a drive mechanism therein, comprising: a soundproof exterior covering at least a head of the robot; A pair of external microphones that are provided at both ear positions corresponding to both ears on the outside of the exterior and mainly collect external sounds, and a noise mainly from an internal noise source provided on the inside of the exterior. At least one internal microphone that collects sound, and a processing circuit that cancels a noise signal from an internal noise source from an acoustic signal from the external microphone based on a signal from the external microphone and the internal microphone, The processing circuit cancels noise of the acoustic signal from the external microphone by performing arithmetic processing on the acoustic signal from the external microphone and the acoustic signal from the internal microphone. The one in which the features.

In the robot hearing device according to the present invention, preferably, the processing circuit previously stores noise generated by a driving mechanism inside the robot, and performs arithmetic processing on a voice signal from an external microphone and a voice signal from the internal microphone. At this time, the noise cancellation processing is performed by referring to the stored noise.

In the robot hearing device according to the present invention, preferably, the processing circuit previously stores a noise characteristic component extracted from noise generated by a driving mechanism inside the robot, and outputs a voice signal from an external microphone and an internal microphone. When performing the arithmetic processing on the audio signal from the CPU, the noise canceling processing is performed by referring to the stored noise characteristic component.

Preferably, in the robot hearing device according to the present invention, when the processing circuit performs arithmetic processing on a voice signal from an external microphone and a voice signal from an internal microphone while learning noise from a driving mechanism inside the robot, The noise cancellation process is performed by referring to the learned noise.

According to the above configuration, the external microphone mainly collects sound from an external target, and the internal microphone mainly collects noise from a noise generating source such as a driving mechanism inside the robot. At that time, the noise signal from the noise source inside the robot was mixed in the acoustic signal collected by the external microphone, and the mixed noise signal was collected by the internal microphone by arithmetic processing in the processing circuit. It is significantly reduced by being canceled by a noise signal. Therefore, in the acoustic signal from the external microphone, the noise from the noise source such as the drive mechanism inside the robot is easily and significantly reduced by the arithmetic processing in the processing circuit, and the S / N ratio is greatly improved. Active perception can be performed even better.

The processing circuit previously stores noise generated by a noise source inside the robot. When processing the audio signal from the external microphone and the audio signal from the internal microphone, the stored noise is stored in the processing circuit. When the noise canceling process is performed by referring to the information, the noise canceling process can be performed more accurately and more quickly because each noise is accurately recorded in advance. .

The above processing circuit has previously stored a noise characteristic component extracted from noise generated by a noise source inside the robot, and performs arithmetic processing on a voice signal from an external microphone and a voice signal from an internal microphone. In the case where the noise canceling process is performed by referring to the stored noise feature component, the storage capacity can be reduced by storing only the noise feature portion.

When the processing circuit performs arithmetic processing on a voice signal from an external microphone and a voice signal from an internal microphone while learning noise from a noise source inside the robot,
When the noise cancellation process is performed by referring to the learned noise, perfect active hearing can always be realized by learning the noise change even if the noise due to the noise source changes.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail based on embodiments shown in the drawings. 1 to 5 show the overall configuration of an experimental humanoid robot provided with an embodiment of the robot hearing device according to the present invention. In FIG. 1, a humanoid robot 10 is configured as a 4DOF (degree of freedom) robot, and includes a base 11 and a base 11.
The body 12 is supported so as to be rotatable around one axis (vertical axis). On the body 12, three body directions (vertical axis,
(A horizontal axis in the left-right direction and a horizontal axis in the front-rear direction).

The base 11 may be fixedly arranged, and may be operable as a leg. Also base 1
1 may be mounted on a movable cart or the like.

The body 12 is rotatably supported on the base 11 about a vertical axis as shown by an arrow A in FIG. 1. The body 12 is rotated by driving means (not shown). In the case shown, it is covered with a soundproof exterior.

The head 13 is supported by the body 12 via a connecting member 13a.
Can swing about a horizontal axis in the front-rear direction, as shown by an arrow B in FIG. 1, and can swing about a horizontal axis in the left-right direction, as shown by an arrow C in FIG. In addition, the connecting member 13a is swingably supported as shown by the arrow D in FIG. It is rotationally driven in the directions of arrows A, B, C, and D by the drive means not provided.

The head 13 is entirely covered with a soundproof exterior 14 as shown in FIG.
A camera 15 is provided on the front side as a visual device in charge of robot vision, and a pair of external microphones 16 (16a, 16b) are provided on both sides as hearing devices in charge of robot hearing. Further, as shown in FIG.
A pair of internal microphones 17 (17a, 17b) are provided inside the exterior 14 so as to be separated from each other on the left and right.

The exterior 14 is made of a sound-absorbing synthetic resin such as urethane resin, for example. The interior of the head 13 is sound-insulated by almost completely sealing the interior of the head 13. Have been. The exterior of the body 12 is similarly made of a sound-absorbing synthetic resin.

The camera 15 has a known configuration, and a commercially available camera having a so-called pan, tilt, and zoom 3DOF (degree of freedom) can be used.

Each of the external microphones 16 has a head 13
Is attached so as to have directivity toward the front side. Here, the left and right external microphones 16a and 16b of the external microphone 16 are attached to the inside by stepped portions 14a and 14b facing forward on both sides of the exterior 14, respectively, as shown in FIGS. Part 14a,
The front sound is collected through the through-hole provided in 14b, and the sound is shielded by appropriate means so as not to pick up the sound inside the exterior 14. Thereby, the external microphone 16
a and 16b are configured as so-called binaural microphones. In addition, the exterior 14 may be formed in the shape of the outer ear of a human near the mounting position of the external microphones 16a and 16b.

Each of the internal microphones 17 is
Inside the external microphones 16a, 16b
, In the case shown in the figure, above the vicinity of both ends of the camera 15. The internal microphone 17 is
It may be provided at an arbitrary position inside the exterior 14.

FIG. 4 shows an electrical configuration including the external microphone 16 and the internal microphone 17 for sound processing.
In FIG. 4, acoustic signals SOL, SOR, S from the external microphones 16a, 16b and the internal microphones 17a, 17b are shown.
IL and SIR are input to processing circuits 18 and 19, respectively. That is, the acoustic signal S from the left external microphone 16a
OL and sound signal SIL from left internal microphone 17a
Is input to the processing circuit 18 and the right external microphone 16
b and the acoustic signal SIR from the left internal microphone 17b are input to the processing circuit 19.

The processing circuits 18 and 19 have the same configuration, and the configuration of the processing circuit 20 will be described below with reference to FIG. 5, the processing circuit 20 includes a filter circuit 21 and a subtraction circuit 22. As the filter circuit 21, a filter circuit having a known configuration, for example, an inverse filter or an adaptive filter is used.
Signal (SIL, SI) from (17a, 17b)
R) is input, and the subtraction circuit 22 outputs the sound signals SO (SOL, SOR) from the external microphones 16a and 16b.
Is subtracted from the signal from the filter circuit 21.

Here, the filter circuit 21 performs a filtering process on the sound signal SI from the internal microphone 17 to thereby filter the noise signal NI corresponding to the noise component NO included in the sound signal SO from the external microphone 16. Reproduce. Then, the subtraction circuit 22 subtracts the noise signal NI from the sound signal SO from the external microphone 16, and thereby the noise from the noise source such as each driving mechanism inside the robot mixed into the sound signal SO from the external microphone 16. The signal NO is almost completely removed, and the true sound signal SL or SR is output.

The humanoid robot 10 according to the embodiment of the present invention is configured as described above, and includes the external microphones 16a and 16a.
The sound from the target b to be collected is collected and noise canceled as follows. First, the external microphones 16a and 16b mainly collect external sounds from a target and output acoustic signals SOL and SOR, respectively. Here, the external microphones 16a and 16b also collect noise from the inside of the robot, but the exterior 14 itself seals the inside of the head 13 and the external microphones 16a and 16b
, Noise is suppressed to a relatively low level.

On the other hand, the internal microphones 17a, 17b
Collects noise mainly from the inside of the robot, for example, noise from a noise generation source such as the operation sound of each drive mechanism and the operation sound of the cooling fan. Here, the internal microphone 1
7a and 17b also collect sound from the outside, but the level is kept relatively low because the exterior 14 seals the inside.

The external microphone 16 thus collected
The sound signals SOL and SOR from the a and 16b and the sound signals SIL and SIR from the internal microphones 17a and 17b are input to processing circuits 18 and 19, respectively. Processing circuit 1
8 and 19, the sound signals SIL and SIR from the internal microphones 17a and 17b are filtered by the filter circuit 20 from the sound signals SOL and SOR from the external microphones 16a and 16b, respectively, and then subtracted by the subtraction circuit 22. By removing noise signals from noise sources inside the robot from the acoustic signals SOL and SOR from the external microphones 16a and 16b, true acoustic signals SL and SR from which noise has been removed are output to the outside.
Then, based on the acoustic signals SL and SR, a control unit (not shown) performs acoustic recognition to perform active perception.

Thus, according to the humanoid robot 10 according to the embodiment of the present invention, the processing circuits 18 and 19
Acoustic signals SOL, SO from external microphones 16a, 16b
R, the acoustic signal SI from the internal microphones 17a, 17b
By performing noise cancellation based on L and SIR, it is possible to obtain a sound signal with a good S / N ratio while directing the directivity directions of the external microphones 16a and 16b to the target by each drive mechanism, and more accurately. Recognition of an acoustic signal can be performed. Therefore, for example, even when the target is moving, the external microphone 16 is driven by each driving mechanism.
The target acoustic signal can be recognized while the direction of the directivities a and 16b always follow the target.

FIG. 6 shows the configuration of a processing circuit in a second embodiment of the robot hearing device according to the present invention.
6, the processing circuit 23 includes a noise sound storage database 24 as a storage unit, an external microphone noise prediction module 25, and a subtraction circuit 22. The noise sound storage database 24 stores sound signals SO and SI collected by the external microphone 16 and the internal microphone 17 for each noise generated from a single sound source in advance as external noise and internal noise data. At the same time, data of external noise and internal noise corresponding to each other are labeled and associated with each other.

The external microphone noise prediction module 25
The input acoustic signal SI from the internal microphone 17 is collated with internal noise data stored in the noise sound storage database 24 to analyze the type of noise included in the acoustic signal SI. This collation is performed by predicting the closest internal noise SI by combining the data of the internal noise. Then, the external microphone noise prediction module 25 predicts an external noise NI that combines data of the corresponding external noise based on the type of the analyzed noise. The subtraction circuit 22 outputs the sound signal S from the external microphone 16.
The external noise NI predicted by the external microphone noise prediction module 25 is subtracted from O.

According to the processing circuit 23 having such a configuration,
On the basis of the sound signal SI from the internal microphone 17, the external microphone noise prediction module 25 refers to the data of the internal noise and the external noise stored in the noise sound storage database 24 to generate the sound signal SO from the external microphone 16. A noise signal NI corresponding to the included noise component NO is predicted. Then, by subtracting the noise signal NI from the sound signal SO from the external microphone 16 in the subtraction circuit 22, the noise signal from the noise generating source such as each drive mechanism inside the robot mixed into the sound signal SO from the external microphone 16 is obtained. The noise signal NO is almost completely removed, and the true acoustic signal SL or SR is output.

FIG. 7 shows the configuration of a processing circuit in a third embodiment of the robot hearing device according to the present invention.
7, the processing circuit 26 has the same configuration as the processing circuit 23 shown in FIG. 6, but the data format of the internal noise and the external noise stored in the noise sound storage database 24 is such that axis,
This is an abstraction (normalization) in the power axis or frequency axis direction, and the external microphone noise prediction module 25
The configuration is different in that a shift process is performed on the input acoustic signal SI from the internal microphone 17 in the time, frequency, and power directions, and then matching is performed.

According to the processing circuit 26 having such a configuration,
The external microphone noise prediction module 25
7 is normalized, and the noise sound storage database 2 is generated based on the normalized sound signal SI.
4, the noise signal NI corresponding to the noise component NO included in the sound signal SO from the external microphone 16 is predicted by referring to the data of the internal noise and the external noise stored in the external microphone 16. The noise signal NI is subtracted from the acoustic signal SO.

As a result, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generating source such as each driving mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output. At the same time, noise data of a plurality of patterns can be combined into one abstract data, so that the size of the database can be reduced.

FIG. 8 shows the configuration of a processing circuit in a fourth embodiment of the robot hearing device according to the present invention.
8, the processing circuit 27 has substantially the same configuration as the processing circuit 23 shown in FIG. 6, and has a different configuration in the following points. That is, the external microphone noise prediction module 25 outputs the input acoustic signal SI from the internal microphone 17.
Is compared with the internal noise data stored in the noise sound storage database 24 to analyze the type of noise included in the acoustic signal SI. At the time of this collation, the external microphone noise module 25 refers to the operation information 40 of the robot 10, that is, the motor operation information indicating which motor of which drive mechanism of the robot 10 has driven and how much noise included in the acoustic signal SI. Analyze the type of

According to the processing circuit 27 having such a configuration,
Based on the acoustic signal SI from the internal microphone 17, the external microphone noise prediction module 25 refers to the internal noise and external noise data stored in the noise sound storage database 24 and the robot operation information 40, and Noise component NO contained in the acoustic signal SO from the
Is predicted. Then, the noise signal NI is subtracted from the sound signal SO from the external microphone 16 by the subtraction circuit 22, so that the noise from the noise source such as each driving mechanism inside the robot mixed into the sound signal SO from the external microphone 16. The signal NO is almost completely removed, and the true sound signal SL or SR is output.

In this case, the external microphone noise module 2
5 can know which motor operation noise is occurring at the time of collation, so that the calculation amount and the computation time required for collation can be reduced.

FIG. 9 shows the configuration of a processing circuit in a fifth embodiment of the robot hearing device according to the present invention.
9, the processing circuit 28 has the same configuration as the processing circuit 27 shown in FIG. 8, but the processing circuit 26 shown in FIG.
Similarly to the above, the data format of the internal noise and the external noise stored in the noise sound storage database 24 is an abstraction (normalization) of a waveform or a power spectrum in a time axis, a power axis, or a frequency axis direction. After the microphone noise prediction module 25 performs a shift process with respect to the time, frequency, and power direction on the input acoustic signal SI from the internal microphone 17,
The configuration is different in that the matching is performed.

According to the processing circuit 28 having such a configuration,
The external microphone noise prediction module 25 is
After normalizing the acoustic signal SI from the noise sound storage database 24 based on the normalized acoustic signal SI.
The noise signal NI corresponding to the noise component NO included in the acoustic signal SO from the external microphone 16 is predicted with reference to the internal noise and external noise data stored in the Then, the noise signal NI is subtracted from the sound signal SO from the external microphone 16. Thus, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generation source such as each drive mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output.

In this case, since the noise data of a plurality of patterns can be combined into one abstract data, the database can be reduced in size, and the external microphone noise module 25 can perform Since it is possible to know which motor is generating the operation noise, it is possible to reduce the amount of calculation and the calculation time required for the verification.

FIG. 10 shows the configuration of a processing circuit in a sixth embodiment of the robot hearing device according to the present invention. 10, the processing circuit 29 has the same configuration as the processing circuit 23 shown in FIG. 6, but instead of the noise sound storage database 24, the noise sound model database 3
0 is provided.

This noise sound model database 30 is
External microphone 1 for each noise generated from a single sound source
6 and the sound signals SO and SI collected by the internal microphone 17
Characteristics extracted from the audio signal, for example, Attack, FM, AM, and Harmonic extracted from the acoustic signal SI.
Features such as s are digitized, and the feature amounts are stored as noise models using distribution expressions such as average values and variance values. Note that the noise model stored in the noise sound model database 30 reduces the number of feature amounts by expressing each feature amount in a rectangular coordinate space using principal component analysis or the like for each noise. May be reduced.

According to the processing circuit 29 having such a configuration,
The external microphone noise prediction module 25
7 with reference to the internal noise model stored in the noise sound model database 30 based on the acoustic signal SI from
Nearest acoustic signal SI by combining internal noise models
Or, conversely, analyze the acoustic signal SI to extract a feature amount, calculate which internal noise model is closer to which internal noise model by using, for example, a least squares method, discriminant analysis, or the like. The type of noise included in the sound signal SO from the microphone 16 is predicted.

Then, the external microphone noise prediction module 25 predicts the noise signal NI by combining the external noise model of the noise sound model database 30 based on the type of the predicted noise, and the subtraction circuit 22 outputs the noise signal NI from the external microphone 16. The predicted noise signal NI is subtracted from the acoustic signal SO. Thus, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generation source such as each drive mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output.

In this case, the noise sound model database 3
0 stores only the noise feature components extracted from the audio signals SO and SI, so that the data amount can be reduced, the entire database can be reduced in size, and the amount of calculation by the external microphone noise prediction module 25 can be reduced. And the calculation time is reduced.

FIG. 11 shows a configuration of a processing circuit in a seventh embodiment of the robot hearing device according to the present invention. In FIG. 11, the processing circuit 31 has substantially the same configuration as the processing circuit 29 shown in FIG. 10, and is different in the following points.

The external microphone noise prediction module 25
The input acoustic signal SI from the internal microphone 17 is compared with the internal noise model stored in the noise sound model database 30 to analyze the type of noise included in the acoustic signal SI. At the time of this collation, the external microphone noise module 25 outputs the operation information 40 of the robot 10, that is, the robot 1.
The type of noise included in the audio signal SI is analyzed with reference to motor operation information indicating which motor of which drive mechanism has been driven.

According to the processing circuit 31 having such a configuration,
The external microphone noise prediction module 25
7, the internal noise model stored in the noise sound model database 30 and the robot operation information 40 are referred to, and the internal noise model is combined to predict the closest acoustic signal SI, or vice versa. The signal SI is analyzed to extract a feature amount, which internal noise model is closer to which internal noise model is calculated by using, for example, a least squares method or discriminant analysis, and the like. Predict the type of noise to be included.

Then, the external microphone noise prediction module 25 predicts the noise signal NI by combining the external noise model of the noise sound model database 30 based on the predicted noise type, and the subtraction circuit 22 outputs the noise signal NI from the external microphone 16. The predicted noise signal NI is subtracted from the acoustic signal SO. Thus, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generation source such as each drive mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output.

In this case, the noise sound model database 3
0 stores only the noise feature components extracted from the audio signals SO and SI, so that the data amount can be reduced, the entire database can be reduced in size, and the amount of calculation by the external microphone noise prediction module 25 can be reduced. And the calculation time is reduced. Further, since the external microphone noise module 25 can recognize which motor operation noise is occurring at the time of matching, it is possible to reduce the amount of calculation and calculation time required for matching.

FIG. 12 shows a configuration of a processing circuit in an eighth embodiment of the robot hearing device according to the present invention. 12, the processing circuit 32 has the same configuration as the processing circuit 23 shown in FIG. 6, but has a different configuration in the following points.

The noise sound storage database 24 stores the sound signal SO for the sound signal SO collected by the external microphone 16 as external noise data.
For the acoustic signal SI collected by the internal microphone 17,
Similar to the noise sound model database 30 of the processing circuit 11, the characteristics of the noise extracted from the audio signal SI for each noise generated from a single sound source in advance, for example, Attack, FM, AM, Ha
Features such as rmonics are digitized, and the features are stored as an internal noise model using a distribution expression such as an average value and a variance value. The corresponding external noise data and the internal noise model are labeled and associated with each other. Have been. In this case, the data of the external noise may be stored as the audio signal SO as it is, as in the processing circuit 23, or may be stored as data after normalizing the audio signal SO as in the processing circuit 26. Is also good.

According to the processing circuit 32 having such a configuration,
The external microphone noise prediction module 25
7, the closest acoustic signal SI is predicted by combining the internal noise models with reference to the internal noise model stored in the noise sound storage database 24 on the basis of the acoustic signal SI from 7 or the acoustic signal SI is analyzed conversely. To extract which feature is closer to which internal noise model, using, for example, a least squares method or discriminant analysis,
The type of noise included in the audio signal SO from the external microphone 16 is predicted.

Then, the external microphone noise prediction module 25 predicts the noise signal NI by combining the external noise data of the noise sound storage database 24 based on the predicted noise type. Signal NI predicted from acoustic signal SO from
Is subtracted. Thus, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generation source such as each drive mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output.

In this case, the noise sound storage database 24
Since only the characteristic components of the noise extracted from the acoustic signal SI with respect to the internal noise are stored, the data amount can be reduced, the entire database can be reduced in size, and the amount of calculation by the external microphone noise prediction module 25 can be reduced. And the calculation time is reduced.

FIG. 13 shows the configuration of a processing circuit in a ninth embodiment of the robot hearing device according to the present invention. In FIG. 13, the processing circuit 33 has substantially the same configuration as the processing circuit 32 shown in FIG. 12, and differs in the following points.

The external microphone noise prediction module 25
The type of the noise included in the acoustic signal SI is analyzed by comparing the input acoustic signal SI from the internal microphone 17 with the internal noise model stored in the noise sound storage database 24. At the time of this collation, the external microphone noise module 25 outputs the operation information 40 of the robot 10, that is, the robot 10
The type of the noise included in the acoustic signal SI is analyzed with reference to the motor operation information indicating which of the driving mechanisms has driven the motor and how much.

According to the processing circuit 33 having such a configuration,
The external microphone noise prediction module 25
7, the closest acoustic signal SI is predicted by combining the internal noise model with reference to the internal noise model stored in the noise sound storage database 24 and the robot operation information 40 based on the acoustic signal SI from The acoustic signal SI is analyzed to extract a feature amount, and which internal noise model is close to the feature amount is calculated using, for example, a least squares method or discriminant analysis, and the acoustic signal SO from the external microphone 16 is calculated. Predict the type of noise included in.

Then, the external microphone noise prediction module 25 predicts the noise signal NI by combining the external noise data of the noise sound storage database 24 based on the predicted noise type. Signal NI predicted from acoustic signal SO from
Is subtracted. Thus, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generation source such as each drive mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output.

In this case, the noise sound storage database 24
Since only the characteristic components of the noise extracted from the acoustic signal SI with respect to the internal noise are stored, the data amount can be reduced, the entire database can be downsized, and the amount of calculation by the external microphone noise prediction module 25 can be reduced. And the calculation time is reduced. Further, since the external microphone noise module 25 can recognize which motor operation noise is occurring at the time of comparison,
The amount of calculation and the calculation time required for collation can be reduced.

FIG. 14 shows a configuration of a processing circuit in a tenth embodiment of the robot hearing device according to the present invention. 14, the processing circuit 34 has substantially the same configuration as the processing circuit 29 shown in FIG. 10, and has a different configuration in that an external microphone noise module 35 is provided instead of the external microphone noise prediction module 25. Has become.

The external microphone noise prediction module 35
Similar to the external microphone noise prediction module 25, the input acoustic signal SI from the internal microphone 17 is compared with the internal noise model stored in the noise sound model database 30 to analyze the type of noise included in the acoustic signal SI. I do. Further, the external microphone noise prediction module 35
The type of noise analyzed is the noise sound model database 3
If the internal noise model is significantly different from the internal noise model stored in 0, it is stored in the noise sound model database 30 as an internal noise model of an unknown noise sound. This allows
An unknown noise sound will be learned.

According to the processing circuit 34 having such a configuration,
The external microphone noise prediction module 35
7 with reference to the internal noise model stored in the noise sound model database 30 based on the acoustic signal SI from
Nearest acoustic signal SI by combining internal noise models
Or, conversely, analyze the acoustic signal SI to extract a feature amount, calculate which internal noise model is closer to which internal noise model by using, for example, a least squares method, discriminant analysis, or the like. The type of noise included in the sound signal SO from the microphone 16 is predicted.

Then, the external microphone noise prediction module 35 predicts the noise signal NI by combining the external noise model of the noise sound model database 30 based on the type of the predicted noise, and the subtraction circuit 22 outputs the noise signal NI from the external microphone 16. Noise signal NI predicted from the acoustic signal SO
Is subtracted. Thus, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generation source such as each drive mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output.

Further, the external microphone noise module 35
If the type of the analyzed noise is significantly different from the internal noise model stored in the noise sound model database 30, the robot surrounds the robot by learning the type of the noise as an internal noise model of an unknown noise sound. It is possible to cope with a change in noise sound due to a change in environment or the like. In this case, since the new internal noise model to be learned is so-called unsupervised learning, it is preferable to use a method of performing self-organization such as a neural network to use a noise model determined to be the same noise sound. In the same category. In this case, not only noise noise due to the motor of the drive mechanism but also noise noise due to other factors, for example, noise such as a rubbing sound of the cover can be removed. In this case, instead of the noise sound model database 30, the noise sound storage database 24 shown in FIG. 6, FIG. 7, FIG. 12 or FIG.
May be used.

FIG. 15 shows the configuration of a processing circuit in the eleventh embodiment of the robot hearing device according to the present invention. In FIG. 15, the processing circuit 36 has substantially the same configuration as the processing circuit 34 shown in FIG. 14, and differs in the following points.

The external microphone noise prediction module 35
The input acoustic signal SI from the internal microphone 17 is collated with the internal noise model stored in the noise sound model database 30 to analyze the type of noise included in the acoustic signal SI. At the time of this collation, the external microphone noise module 35 outputs the operation information 40 of the robot 10, that is, the robot 1.
The type of noise included in the acoustic signal SI is analyzed with reference to the motor operation information indicating which drive mechanism of which drive mechanism has been driven. Further, when the type of the analyzed noise is significantly different from the noise model stored in the noise sound model database 30, the external microphone noise prediction module 35 sets the robot motion information as an internal noise model of an unknown noise sound. The data is stored in the noise sound model database 30 in association with 40. Thereby, an unknown noise sound is learned.

According to the processing circuit 36 having such a configuration,
The external microphone noise prediction module 35
7, the closest acoustic signal SI is predicted by combining the internal noise model with reference to the internal noise model stored in the noise sound model database 30 and the robot operation information 40 based on the acoustic signal SI from The acoustic signal SI is analyzed to extract a feature amount, and which internal noise model is close to the feature amount is calculated using, for example, a least squares method or discriminant analysis, and the acoustic signal SO from the external microphone 16 is calculated. Predict the type of noise included in.

Then, the external microphone noise prediction module 35 predicts the noise signal NI by combining the external noise model of the noise sound model database 30 based on the predicted noise type, and the subtraction circuit 22 outputs the noise signal NI from the external microphone 16. The predicted noise signal NI is subtracted from the acoustic signal SO. Thus, the noise signal NO mixed with the sound signal SO from the external microphone 16 from the noise generation source such as each drive mechanism inside the robot is almost completely removed, and the true sound signal SL or SR is output. Further, since the external microphone noise prediction module 35 can recognize which motor operation noise is occurring at the time of matching, it is possible to reduce the amount of calculation and calculation time required for matching, and Since a new internal noise model is learned, it is possible to cope with a change in noise sound due to a change in the environment surrounding the robot.

In the above-described embodiment, the humanoid robot 10 is configured to have 4 DOF (degree of freedom). However, the present invention is not limited to this, and the robot according to the present invention may be configured to perform any operation. It is also possible to incorporate a robot hearing device. In the above-described embodiment, the robot auditory device according to the present invention is connected to the humanoid robot 1.
Although the description has been given of the case where the robot is incorporated in the robot 0, it is apparent that the invention is not limited to this, and that the robot can be incorporated in various animal-type robots such as a dog-type robot and other types of robots. Further, in the above-described embodiment, the internal microphone 17 is constituted by a pair of microphones 17a and 17b, but may be a single microphone, or may be, for example, a large number of microphones provided for each noise source. You may. Similarly, it is clear that the external microphone 16 may be a single microphone, or may be composed of three or more microphones.

[0075]

As described above, according to the present invention,
The external microphone mainly collects sound from an external target, and the internal microphone mainly collects noise from a noise source such as a drive mechanism inside the robot. At this time, the noise signal from the noise source inside the robot is mixed in the acoustic signal collected by the external microphone, and the mixed noise signal is the noise signal collected by the internal microphone by the arithmetic processing in the processing circuit. And is significantly reduced. Therefore, as for the acoustic signal from the external microphone, the noise signal from the noise source such as the driving mechanism inside the robot is remarkably reduced, so that the S / N ratio is greatly improved, and the active perception is further improved. Can do it. Thus, according to the present invention, it is possible to perform an active perception by collecting sounds from an external target without being affected by noise generated inside the robot such as a driving mechanism. A robot hearing device is provided.

[Brief description of the drawings]

FIG. 1 is a front view showing the appearance of a humanoid robot incorporating a first embodiment of a robot hearing device according to the present invention.

FIG. 2 is a side view of the humanoid robot of FIG. 1;

FIG. 3 is a schematic enlarged view showing a configuration of a head in the humanoid robot of FIG. 1;

FIG. 4 is a block diagram showing an electrical configuration of a robot hearing device in the humanoid robot of FIG. 1;

FIG. 5 is a schematic diagram showing a configuration of a processing circuit in the robot hearing device of FIG. 4;

FIG. 6 is a schematic diagram showing a configuration of a processing circuit in a second embodiment of the robot hearing device according to the present invention.

FIG. 7 is a schematic diagram showing a configuration of a processing circuit in a third embodiment of the robot hearing device according to the present invention.

FIG. 8 is a schematic diagram showing a configuration of a processing circuit in a fourth embodiment of the robot hearing device according to the present invention.

FIG. 9 is a schematic diagram illustrating a configuration of a processing circuit in a fifth embodiment of the robot hearing device according to the present invention.

FIG. 10 is a schematic diagram showing a configuration of a processing circuit in a sixth embodiment of the robot hearing device according to the present invention.

FIG. 11 is a schematic diagram illustrating a configuration of a processing circuit in a seventh embodiment of the robot hearing device according to the present invention.

FIG. 12 is a schematic diagram showing a configuration of a processing circuit in an eighth embodiment of the robot hearing device according to the present invention.

FIG. 13 is a schematic diagram showing a configuration of a processing circuit in a ninth embodiment of the robot hearing device according to the present invention.

FIG. 14 is a schematic diagram showing a configuration of a processing circuit in a tenth embodiment of the robot hearing device according to the present invention.

FIG. 15 is a schematic diagram showing a configuration of a processing circuit in an eleventh embodiment of the robot hearing device according to the present invention.

[Explanation of symbols]

10 Humanoid robot 11 Base 12 Body 13 Head 13a Connecting member 14 Exterior 15 Camera (robot perception) 16, 16a, 16b External microphone 17, 17a, 17b Internal microphone 18, 19, 20, 23, 26, 27, 28 , 29,3
1, 32, 33, 34, 35 processing circuit 21 filter circuit 22 subtraction circuit 24 noise sound storage database 25 external microphone noise prediction module 30 noise sound model database

Claims (5)

[Claims] 1. A robot provided with a noise source inside, comprising: a soundproof exterior covering at least a part of the robot; and at least one external microphone provided outside the exterior and mainly collecting external sounds. And at least one internal microphone that is provided inside the exterior and collects noise mainly from an internal noise source, based on signals from the external microphone and the internal microphone,
And a processing circuit for canceling a noise signal from an internal noise source from an audio signal from an external microphone. A robot auditory apparatus characterized by canceling noise of an acoustic signal from an external microphone. 2. A humanoid or animal type robot provided with a noise generating source such as a drive mechanism therein, wherein a soundproof exterior covering at least the head of the robot and ears on both sides outside the exterior are provided. A pair of external microphones that are provided at both ear positions to collect mainly external sound, and at least one internal microphone that is provided inside the exterior and mainly collects noise from an internal noise source. , Based on the signals from the external microphone and the internal microphone,
And a processing circuit for canceling a noise signal from an internal noise source from an audio signal from an external microphone. A robot auditory apparatus characterized by canceling noise of an acoustic signal from an external microphone. 3. The processing circuit has previously stored noise generated by a driving mechanism inside the robot, and has the stored noise when performing arithmetic processing on a voice signal from an external microphone and a voice signal from an internal microphone. 3. The robot hearing device according to claim 1, wherein the noise canceling process is performed by referring to. 4. The processing circuit stores in advance noise characteristic components extracted from noise generated by a driving mechanism inside the robot, and performs arithmetic processing on a voice signal from an external microphone and a voice signal from an internal microphone. 3. The robot hearing device according to claim 1, wherein the noise canceling process is performed by referring to the stored noise feature component. 5. The processing circuit according to claim 1, wherein the arithmetic circuit processes an audio signal from an external microphone and an audio signal from an internal microphone while learning noise caused by a driving mechanism inside the robot.
3. A noise canceling process is performed by referring to the learned noise.
Robotic hearing device.