JP2022057414A - Sound source position estimation device, sound source position estimation method and sound source position estimation program - Google Patents

Abstract

To provide a sound source position estimation device, a sound source position estimation method and a sound source position estimation program which facilitate estimation of a sound source direction and a sound source position of a true survey object sound source by performing proper processing for many obtained estimation directions.SOLUTION: Sound source position estimation devices 1, 21 for estimating positions U1 (x, y)...of survey object sound sources U1... by using reception signals R1 (t)... obtained from ultrasonic sensors S1... arranged on a first axial line X comprise: a starting point deflection angle acquisition unit 13 which obtains estimation deflection angles βmn1... heading toward estimation sound sources E1... estimated to be a survey object sound source by using an intermediate point between a pair of ultrasonic sensors Sm, Sn as a starting point Kmn using a pair of reception signals Rm(t), Rn(t); an estimation direction line generation unit 14 which generates estimation direction lines Lmn1... extending at the estimation deflection angles βmn1... from each starting point Kmn in an effective XY region RXY; and direction line selection processing units 15, 25 which select a true estimation direction line Lmnt or exclude an untrue estimation direction line Lmnf.SELECTED DRAWING: Figure 3

Description

The present invention relates to a sound source position estimation device that estimates the position of a sound source to be explored using a received signal of an ultrasonic sensor, a sound source position estimation method, and a sound source position estimation program.

Conventionally, a sound source position estimation device that estimates the direction and position of a sound source that emits ultrasonic waves and a pseudo sound source (reflector, obstacle) that reflects ultrasonic waves using multiple received signals obtained from each of multiple ultrasonic sensors. And obstacle detection systems are known (see, for example, Patent Document 1).
Further, as a method for estimating the direction of a sound source, for example, a method described in Patent Document 2 (see FIG. 2) is known.

Japanese Unexamined Patent Publication No. 2012-127863 Japanese Patent No. 6413741

For example, using received signals received by three or more ultrasonic sensors arranged in a row, a cross-correlation function is calculated from the received signals of paired ultrasonic sensors, and one or more peak time differences are obtained. An estimation direction of one or a plurality of sound sources or pseudo sound sources (hereinafter referred to as “explored sound sources”) starting from an intermediate point of a pair of ultrasonic sensors is obtained. This may be performed for each combination of paired ultrasonic sensors, and the direction (argument) and position of each sound source to be explored may be estimated using a large number of acquired estimation directions.

However, when there are two or more sound sources to be explored, the estimation direction obtained from the pair of received signals of the paired ultrasonic sensors includes the true estimation direction indicating the direction of the genuine sound source to be searched. Also includes an untrue estimation direction indicating the direction of a false sound source (false image) due to the cross-correlation function obtained for signals from different sound sources to be explored contained in a pair of received signals.

However, it was not possible to properly determine which of the many estimation directions obtained was the true estimation direction and which was the inauthentic estimation direction.
The present invention has been made in view of the above problems, and is a sound source that facilitates estimation of the sound source direction and sound source position of a genuine sound source to be explored by performing appropriate processing on a large number of obtained estimation directions. A position estimation device, a sound source position estimation method, and a sound source position estimation program are provided.

(1) In one embodiment, P (P is) using received signals obtained from N (N is a natural number of 3 or more) ultrasonic sensors arranged at intervals on the first axis. An intermediate point between the pair of ultrasonic sensors using the pair of received signals obtained from the pair of ultrasonic sensors in a sound source position estimation device that estimates the position of the probed sound source (1 or more natural numbers). The starting point declination acquisition unit that acquires the estimated declination from the starting point toward the estimated sound source estimated to be the probed sound source for each combination of the ultrasonic sensors, the first axis line, and the first axis. Of the virtual XY plane formed by the second axis extending from the first axis in the direction orthogonal to the first axis, the effective XY region where the position of the explored sound source can be effectively estimated is declined from each starting point. Of the estimation direction line generator that generates the estimation direction line that forms an angle and extends toward the estimation sound source and the estimation direction line generated in the effective XY region, the authentic estimation direction line toward the probed sound source is genuine. A sound source including a direction line selection processing unit that selects an estimated direction line presumed to be It is a position estimation device.

(2) In another embodiment, P (P) (P) are obtained by using received signals obtained from N (N is a natural number of 3 or more) ultrasonic sensors arranged at intervals on the first axis. Is a method of estimating the sound source position for estimating the position of the probed sound source (of 1 or more natural numbers), and is used between the pair of ultrasonic sensors using the pair of received signals obtained from the pair of the ultrasonic sensors. The starting point declination acquisition step for acquiring the estimated declination from the starting point toward the estimated sound source estimated to be the probed sound source for each combination of the ultrasonic sensors, starting from the intermediate point, and the first axis. Of the virtual XY plane formed by the second axis extending from the first axis in the direction orthogonal to the first axis, the effective XY region in which the position of the explored sound source can be effectively estimated is located in the effective XY region from each of the above starting points. Of the estimation direction line generation step of generating the estimation direction line extending toward the estimated sound source by making the estimation declination and the estimation direction line generated in the effective XY region, the authenticity estimation toward the probed sound source is genuine. It comprises a direction line selection step of selecting an estimated direction line presumed to be a direction line, or excluding an estimated direction line presumed to be an inauthentic estimated direction line toward the above-mentioned estimated sound source of inauthency. This is a sound source position estimation method.

(3) In yet another embodiment, P (P) (P) are obtained by using received signals obtained from N ultrasonic sensors (N is a natural number of 3 or more) arranged at intervals on the first axis. Is a sound source position estimation program that estimates the position of the probed sound source (of 1 or more natural numbers), and the computer uses the pair of received signals obtained from the pair of ultrasonic sensors to obtain the pair of ultrasonic sensors. The starting point declination acquisition unit that acquires the estimated declination from the starting point toward the estimated sound source presumed to be the probed sound source for each combination of the ultrasonic sensors, starting from the intermediate point between them, and the first Of the virtual XY plane formed by the axis and the second axis extending in the direction orthogonal to the first axis from the first axis, the above-mentioned effective XY regions in which the position of the explored sound source can be effectively estimated Of the estimated direction line generator that generates the estimated direction line extending from the starting point toward the estimated sound source and the estimated direction line generated in the effective XY region, the direction is toward the genuine probed sound source. With a direction line selection processing unit that selects an estimated direction line that is presumed to be a genuine estimated direction line, or excludes an estimated direction line that is presumed to be an untrue estimated direction line toward the above-mentioned estimated sound source of inauthency. It is a sound source position estimation program that works.

In this device, the starting point deviation angle acquisition unit uses a pair of received signals Rm (t) and Rn (t) of a pair of ultrasonic sensors Sm and Sn, and the intermediate point between the pair of ultrasonic sensors is set as the starting point Kmn. , Estimated deviation angles _βmn1 , _βmn2 , ... Toward the estimated sound sources E1, E2, ... And the estimated deviation angles βmn1, βmn2, ... Are acquired in large numbers.

Further, in this apparatus, in the estimation direction line generation unit, a large number of previously acquired starting points Kmn and estimated declinations βmn1, βmn2, ... Are used for the effective XY region RXY of the virtual XY plane PXY, and each starting point is used. Estimated declinations βmn1, βmn2, ..., Are formed from Kmn to generate estimated direction lines Lmn1, Lmn2, ... Extending toward the estimated sound sources E1, E2, ....
Then, in the direction line selection processing unit, it is estimated that among the estimated direction lines Lmn1, Lmn2, ... Generated in the effective XY region RXY, the genuine estimated direction lines U1, ... The estimation direction line is selected, or the estimation direction line estimated to be the inauthenticity estimation direction line Lmnf toward the inauthentic estimated sound source Ef is excluded.

As described above, in this apparatus, since the estimated direction lines Lmn1, Lmn2, ... It becomes easy to understand the mutual relationship between. Then, the estimated direction line estimated to be the genuine estimated direction line Lmnt is selected, or the estimated direction line estimated to be the untrue estimated direction line Lmnf is excluded, so that the sound source direction of the genuine explored sound source is obtained. And the sound source position can be easily estimated.
This advantage is the same in the above-mentioned method and also in the computer functioning as each processing unit in the program.

In this case, "ultrasonic wave" means a sound used for a purpose other than human hearing, and it does not matter whether it is audible to humans or not.
In addition, the "explored sound wave" refers to a sound wave that emits ultrasonic waves toward an ultrasonic sensor, which is explored by a sound source position estimation device, and emits ultrasonic waves from a sound wave that emits ultrasonic waves by itself or from an ultrasonic sound wave that exists separately. A pseudo sound source (reflector) that becomes a pseudo (secondary) sound wave by reflecting the generated ultrasonic waves is also included.

Further, when a pair of received signals Rm (t) and Rn (t) obtained from the pair of ultrasonic sensors Sm and Sn are used (calculating the cross-correlation function Cmn (τ) from these two received signals), P. When there are multiple explored sound sources U1, ..., UP, the cross-correlation function Cmn (τ) shows a local peak time difference τ (received time difference τmn1, τmn2) according to the arrangement of each explored sound source. , ...,) can be obtained up to ² Ps. The estimated declination βmn1 and the like can be obtained from these wave reception time differences τmn1 and the like. In addition, in the combination of different ultrasonic sensors, the number of obtained estimated declination βmn1 and the like may be different. The estimated declination βmn1 and the like are the declination from the first axis (X-axis) (the vector toward each estimated sound source E1 and the like from the starting point Kmn and the first one, which indicate the estimation direction from the starting point Kmn to each estimated sound source E1 and the like. The declination formed by the axis (X-axis).

Further, in the direction line selection processing unit, an effective XY area is used when selecting an estimated direction line estimated to be a genuine estimated direction line or excluding an estimated direction line estimated to be an inauthentic estimated direction line. It is advisable to determine the estimated direction lines to be selected or excluded based on the distribution situation of a large number of estimated direction lines generated in (for example, the difference in the density of the estimated direction line distribution).
Specifically, there is a determination method in which the estimation direction line passing through the region where the estimation direction line distribution is coarse is estimated to be an inauthentic estimation direction line and excluded. Further, there is also a determination method in which an estimated direction line passing only through a region where the estimated direction line distribution is dense is estimated to be a genuine estimated direction line and selected. Furthermore, prior to the selection of the estimated direction line or the exclusion of the estimated direction line, the estimated direction line distribution is adjusted according to the distance from the average position of the ultrasonic sensor in order to facilitate the determination of the estimated direction line to be selected or excluded. It is also possible to perform pre-processing such as correction.

(4) Further, in the sound source position estimation device of (1), the direction line selection processing unit includes a division processing unit that divides the effective XY region into a large number of grid regions having a predetermined grid spacing, and each grid. The counting processing unit that counts the number of the estimated direction lines passing through the region and the estimated direction line passing through the grid region in which the number of the estimated direction lines is less than the threshold number are the false estimation direction lines. It is preferable to use a sound source position estimation device having an exclusion processing unit that is presumed to exist and excluded.

(5) Further, in the sound source position estimation method of (2), the direction line selection step includes a division step of dividing the effective XY region into a large number of grid regions having a predetermined grid spacing, and within each grid region. It is estimated that the counting step for counting the number of the estimated direction lines passing through the grid and the estimated direction line passing through the grid region where the number of the estimated direction lines is less than the threshold is the false estimation direction line. It is preferable to use a sound source position estimation method having an exclusion step of exclusion.

(6) In addition, in the sound source position estimation program of (3), in order to make the computer function as the direction line selection processing unit, the effective XY region is divided into a large number of grid regions having a predetermined grid spacing. A division processing unit, a counting processing unit that counts the number of the estimated direction lines passing through each grid region, and an estimated direction line passing through the grid region in which the number of the estimated direction lines is less than the threshold number. It is preferable to use a sound source position estimation program that functions as an exclusion processing unit that estimates and excludes the false estimation direction line.

In this sound source position estimation device, the direction line selection processing unit has the above-mentioned division processing unit and the counting processing unit, divides the effective XY region into a large number of grid regions RLxy, and estimates through each grid region RLxy. Since the number Qxy of the direction lines is counted, the coarse and dense state of a large number of estimated direction line distributions can be easily understood as the number Qxy associated with each lattice region RLxy.

Further, the direction line selection processing unit excludes the estimation direction line passing through the lattice region in which the number Qxy is less than the threshold QTh in the exclusion processing unit. Since many genuine estimation direction lines are directed to the genuine sound source to be searched, if a certain estimation direction line is a genuine estimation direction line, the estimated direction line is in the lattice region having a low number Qxy. It is thought that it will not pass. On the contrary, it can be estimated that the estimation direction line passing through the lattice region of the number Qxy less than the threshold number QTh is not the true estimation direction line. Thus, the genuine estimated direction line can be easily selected.
These advantages are the same in the above-mentioned method and also in the computer functioning as each processing unit in the program.

In the sound source position estimation device of (7) or (1), the direction line selection processing unit includes a division processing unit that divides the effective XY region into a large number of grid regions having a predetermined grid spacing, and each grid. By a counting processing unit that counts the number of the estimated direction lines passing through the region, and by weighting correction that the number becomes heavier as the distance from the average position of the N ultrasonic sensors on the XY plane to the grid region increases. , The weighting processing unit that corrects to the number of corrections, and the estimation direction line that passes through the grid region where the number of corrections exceeds the threshold correction number among the estimation direction lines is estimated to be the genuine estimation direction line. It is preferable to use a sound source position estimation device having a selection processing unit for selection.

In the sound source position estimation method of (8) or (2), the direction line selection step includes a division step of dividing the effective XY region into a large number of grid regions having a predetermined grid spacing, and within each grid region. The number of corrections is made by a counting step for counting the number of the estimated direction lines passing through the above and a weighting correction in which the number becomes heavier as the distance from the average position of the N ultrasonic sensors on the XY plane to the grid region increases. A weighting step for correcting to It is preferable to use a sound source position estimation method having.

In the sound source position estimation program of (9) or (3), in order to make the computer function as the direction line selection processing unit, the effective XY region is divided into a large number of grid regions having a predetermined grid spacing. The division processing unit, the counting processing unit that counts the number of the estimated direction lines passing through each grid region, and the number of lines are the distances from the average position of the N ultrasonic sensors in the XY plane to the grid region. The weighting processing unit that corrects to the number of corrections by the weighting correction that becomes heavier as the size increases, and the estimation direction line that passes through the grid region in which the number of corrections exceeds the threshold correction number among the estimation direction lines is the authentic estimation direction. It is preferable to use a selection processing unit that estimates and selects a line and a sound source position estimation program that functions as a line.

Also in this sound source position estimation device, the direction line selection processing unit has the above-mentioned division processing unit and the counting processing unit, divides the effective XY region into a large number of grid regions RLxy, and estimates passing through each grid region RLxy. Since the number Qxy of the direction lines is counted, the coarse and dense state of a large number of estimated direction line distributions can be easily understood as the number Qxy associated with the lattice region RLxy.

Further, the direction line selection processing unit corrects the number Qxy to the corrected number CQxy by a weighting correction that becomes heavier as the distance Dxy from the average position to the grid region RLxy increases. The grid region RLxy near the average position is close to each starting point Kmn, and in many cases, both the true estimation direction line and the inauthentic estimation direction line pass through the same grid region RLxy. On the other hand, in the range away from the average position (distance Dxy is large), there are few cases where both the true estimation direction line and the inauthentic estimation direction line pass through the same grid region RLxy, and only the genuine estimation direction line. Is often passed, or only the estimated inauthentic direction line is passed. Moreover, since many authentic estimation direction lines are directed to the genuine sound source to be explored, even in the lattice region RLxy far from the average position, if the lattice region RLxy through which many authentic estimation direction lines pass, the number Qxy maintains a large value. .. That is, if the grid region RLxy is located in a range where the distance Dxy is large but many estimation direction lines pass through, the estimation direction line passing through the grid region is the true estimation direction line Lmnt. Can be estimated. Therefore, the true estimated direction line Lmnt can be easily selected by selecting the estimated direction line passing through the lattice region where the corrected number CQxy exceeds the threshold correction number CQTh.
These advantages are the same in the above-mentioned method and also in the computer functioning as each processing unit in the program.

As a correction method of the weighting processing unit that performs weighting correction that becomes heavier as the distance Dxy increases, for example, the distance Dxy (= (x ² + y ² ) ^1/2 ) from the average position (origin = (0,0)) The correction coefficient Cxy = C · (x ² + y ² ) ^1/2 proportional to is multiplied by the number Qxy to obtain the correction number CQxy = C · (x ² + y ² ) ^1/2 · Qxy. In addition, as the correction coefficient Cxy that becomes heavier as the distance Dxy increases, it is preferable to determine the correction coefficient Cxy so that the true estimation direction line and the inauthentic estimation direction line can be appropriately distinguished by the threshold correction number CQTh. Depending on the situation of the line distribution, for example, a correction coefficient Cxy = C · (x ² + y ² ) proportional to the square of the distance Dxy may be used, and a correction by multiplying this by the number Qxy may be adopted.

(10) Further, the sound source position estimation device according to any one of (1), (4), and (7), wherein the N ultrasonic sensors have the first position centered on the average position in the XY plane. Estimated that the frequency indicates a local peak in the frequency distribution of the estimated declination of the estimated direction line that is symmetrically arranged on one axis and is selected or excluded by the direction line selection processing unit. It is preferable to use a sound source position estimation device including a declination acquisition unit in which the declination value is the value of the declination toward the probed sound source from the average position.

(11) Further, the sound source position estimation method according to any one of (2), (5), and (8), wherein the N ultrasonic sensors are the first, centering on the average position in the XY plane. Estimated that the frequency indicates a local peak in the frequency distribution of the estimated declination magnitude of the estimated argument line that is symmetrically arranged on one axis and is selected or excluded in the direction line selection step. It is preferable to use a sound source position estimation method including a declination acquisition step in which the declination value is set as the declination value from the average position toward the probed sound source.

(12) Further, the sound source position estimation program according to any one of (3), (6), and (9).
The N ultrasonic sensors are arranged symmetrically on the first axis with the average position in the XY plane as the center, and the computer is selected or excluded from the direction line selection processing unit. Declination acquisition in which the value of the estimated declination at which the frequency indicates a local peak is the value of the declination from the average position toward the sound source to be explored in the frequency distribution of the magnitude of the estimated declination of the estimated direction line. It is good to use a sound source position estimation program that also functions as a unit.

In this sound source position estimation device, the received signals of N ultrasonic sensors arranged symmetrically on the first axis with the average position as the center are used. Then, in the frequency distribution of the magnitude of the estimated declination of the estimated declination selected or excluded by the above-mentioned direction line selection processing unit, the value of the estimated declination at which the frequency indicates a local peak is applied from the average position. It also has a declination acquisition unit for the values of declinations α1, ..., αP toward the exploration sound source. Therefore, it is possible to easily obtain the values of the declinations α1, ..., αP from the average position toward the sound source to be explored.
These advantages are the same in the above-mentioned method and also in the computer functioning as each processing unit in the program.

It is a block diagram which shows the schematic structure of the ultrasonic system for vehicle which comprises a transmitter, an ultrasonic sensor, a sound source position estimation device which concerns on embodiment and a modified form, a display part, and a voice part. A plurality of ultrasonic waves radiated from a transmitter TM arranged at the origin OR00 and reflected by a probed sound wave U separated in the direction of the second axis Y are arranged at intervals on the first axis X. It is explanatory drawing which shows the state of receiving a wave by a sensor. It is a flowchart which shows the processing step performed by the sound source position estimation apparatus which concerns on embodiment. It is explanatory drawing which shows the state that the ultrasonic wave SG from the exploration sound source U reaches the ultrasonic wave sensor Sm, Sn. When two explored sound sources U1 and U2 are present, the received signals Rn (t) and Rn (t) obtained by ultrasonic sensors Sm and Sn and the cross-correlation function Cmn (τ) obtained by using them are obtained. In the explanatory diagram shown, the column (a) is the received signal Rm (t), the column (b) is the received signal Rn (t), and the column (c) is the cross-correlation function Cmn (τ). A direction line layout diagram showing the arrangement state of a large number of estimated direction lines Lmn generated under the condition that two sound sources U1 and U2 to be explored are arranged in the region including the effective XY region near the origin in the XY plane PXY. Is. According to the embodiment, among the estimated direction lines shown in the direction line layout diagram shown in FIG. 6, when the process of removing the estimated direction lines passing through the region where the density of the estimated direction lines is low is performed, a large number of residual authenticities are performed. It is a direction line arrangement diagram which shows the arrangement state of the estimated direction line Lmnt. It is a frequency distribution diagram which shows the relationship between the estimated declination βmn of a large number of residual authenticity estimation direction lines Lmnt and the frequency thereof. Contour processing is performed to obtain a weighted density distribution in which the density of the estimated direction lines in the direction line layout diagram shown in FIG. 6 is weighted by the distance from the origin OR00, and the regions having the same weighted density are connected to each other. It is a distribution map shown. Arrangement state of a large number of selected authentic estimation direction lines Lmnt when processing is performed to select an estimation direction line passing through a region showing a value equal to or larger than the number of threshold corrections in the distribution map of FIG. It is a direction line layout diagram which shows.

(Embodiment)
Hereinafter, the configuration of the ultrasonic system SYS including the sound source position estimation device (hereinafter, also simply referred to as a device) 1 according to the embodiment of the present invention, and the program for functioning the control unit 10 configured by the computer in the device 1 as each unit. PG1 and the processing method in each part will be described with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an ultrasonic system SYS including a sound source position estimation device 1 according to the present embodiment, and FIG. 3 is a flowchart showing a flow of each process in the present embodiment.

The ultrasonic system SYS shown in FIG. 1 is mounted on a vehicle (not shown). This ultrasonic system SYS communicates with a transmitter TM, ultrasonic sensors S1 to SN which are a plurality of receivers, a sound source position estimation device 1, an interface unit 18 of the sound source position estimation device 1, and a CAN bus CB. It is composed of a possible display unit 40 and a voice unit 50 that emits a sound that generates a warning sound or a synthetic voice.

On the front surface of the vehicle (not shown), the first axis X (the axis extending in the vertical direction in FIG. 1) corresponding to the vehicle width direction of the vehicle is arranged at the origin OR00, and the first axis X to the first axis X. Ultrasonic SG that forms a tone burst wave along the second axis Y extending in the direction orthogonal to the one axis X (in the present embodiment, the front of the front-rear direction orthogonal to the width direction of the vehicle; the left direction in FIG. 1). A transmitter TM that emits transmission ultrasonic SGT) is arranged. Further, it is provided with a plurality of ultrasonic sensors S1 to SN (= S8) arranged at intervals on the first axis X and receiving ultrasonic waves SG (received ultrasonic SGR). In this embodiment, SN = 8, and the ultrasonic sensors S1 to S8 (= SN) are equidistant (0.15 m interval) on the first axis X and symmetrical with respect to the origin OR00. The average position of each ultrasonic sensor S1 or the like coincides with the origin OR00. However, the transmitter TM may not be arranged at the origin OR00, the ultrasonic sensors S1 to SN may not be arranged at equal intervals, and may not be arranged symmetrically with respect to the average position.

As a result, as shown in FIG. 2, an object that reflects ultrasonic waves (human beings, obstacles, etc.) is radiated from the transmitter TM as an exploration sound source U that emits ultrasonic waves in a pseudo manner, and the exploration sound source U is emitted. The ultrasonic SG (received ultrasonic SGR) reflected by the above can be received by each ultrasonic sensor S1 to SN.
Of the XY plane PXY (corresponding to the horizontal plane in this embodiment), the position U (x, y) and the direction (argument α from the first axis X at the origin OR00) of the sound source U to be explored are effectively used. The range that can be estimated is defined as the effective XY region RXY. The effective XY region RXY in the present embodiment has an ultrasonic SG (tone burst signal) cycle of 50 msec, a burst duration of 0.2 msec, and an atmospheric atmosphere, which is emitted from the transmitter TM and received by each ultrasonic sensor S1 to SN. Considering the sound velocity V and the like, the range is generally X = ± 4.0 m for the first axis X and Y = 1.0 to 8.0 m for the second axis Y.

In the device 1, in addition to the control unit 10 that performs various controls, the oscillation circuit 2 that drives the transmitter TM to radiate the transmitted ultrasonic SGT from the transmitter TM, and the ultrasonic wave that receives the received ultrasonic SGR. A receiving circuit 3 that arranges signals from sensors S1 to S8 and outputs them as a received signal R1 (t) or the like to a control unit 10 is provided.

Of these, although not shown, the control unit 10 is equipped with general computer hardware such as a CPU (Central Processing Unit), RAM (Random Access Memory), and ROM (Read Only Memory), and the ROM has. , OS (Operating System), sound source position estimation program PG1, various data, and the like are stored. The OS and program PG1 are expanded in RAM and executed by the CPU.

The control unit 10 may be connected to a plurality of computers mounted on the vehicle, such as a CAN bus CB. In FIG. 2, the inside of the control unit 10 shows the function realized by the sound source position estimation program PG1 as a block.

That is, in the present embodiment, the control unit 10 includes the oscillation control unit 11, the signal receiving unit 12, the starting point declination acquisition unit 13, the estimation direction line generation unit 14, the direction line selection processing unit 15, and the sound source declination acquisition unit to be searched. 16. It functions as an explored sound source position acquisition unit 17 and an interface unit 18, and estimates the declination α and the position of the explored sound source U.
The direction line selection processing unit 15 includes a division processing unit 15a, a counting processing unit 15b, and an exclusion processing unit 15c.

In addition, in the present embodiment, each process is performed according to the procedure (step ST1 to step ST7) shown in the flowchart of FIG. 3, and the declination α and the position of the sound source U to be explored are estimated.

Of the control units 10, the oscillation control unit 11 controls the radiation of the transmitted ultrasonic SGT from the transmitter TM through the oscillation circuit 2 (ultrasonic radiation step ST1). The signal receiving unit 12 takes in the received ultrasonic SGR received by the ultrasonic sensor S1 or the like as the received signal R1 (t) or the like through the receiving circuit 3 (reception step ST2).

Further, among the control units 10, the starting point declination acquisition unit 13 estimates the direction of the sound source U to be searched by using the received signal R1 (t) or the like (starting point declination acquisition step ST3). Specifically, in this embodiment, as shown in FIG. 4, a received signal Rm (t) corresponding to a received ultrasonic wave SGR received by a pair of ultrasonic sensors Sm and Sn selected from each ultrasonic sensor S1 and the like. Using the received wave reception time difference τmn generated between Rn (t), the intermediate point between the pair of ultrasonic sensors Sm and Sn is set as the starting point Kmn, and the estimated deviation angle βmn in the direction from this starting point Kmn toward the estimated sound wave E is set. Acquire (1 ≦ m <n ≦ N, m, n are natural numbers). This is acquired for each combination of ultrasonic sensors S1 to _SN (number of combinations NC ₂ , ₈ C ₂ = 28 ways in this embodiment).

This will be described in more detail. First, as a premise, the distance from the explored sound wave U (estimated sound wave E) to the pair of ultrasonic sensors Sm and Sn is sufficiently long, and the received ultrasonic SGRs incident on the pair of ultrasonic sensors Sm and Sn are parallel to each other. do. Further, the position Sm (xm, 0) of the ultrasonic sensor Sm and the position Sn (xn, 0) of the ultrasonic sensor Sn are set. Then, the distance between the ultrasonic sensors Sm and Sn is given by xn-xm. It is assumed that a received ultrasonic wave SGR radiated from the sound source U (estimated sound source E) reaches the ultrasonic sensors Sm and Sn, respectively, but a wave receiving time difference τmn occurs. Then, as shown in FIG. 4, the stroke difference between the stroke from the explored sound source U (estimated sound source E) to the ultrasonic sensor Sm and the stroke from the explored sound source U (estimated sound source E) to the ultrasonic sensor Sn (estimated sound source E). xn−xm) cosβmn is equal to V · τmn obtained by multiplying the received wave time difference τmn by the sound wave V ((xn−xm) cosβmn = V · τmn).
The estimated declination βmn is the angle formed by the first axis X and the received ultrasonic SGR arriving at the ultrasonic sensors Sm and Sn, that is, the starting point Kmn (starting point Kmn) which is an intermediate point between the ultrasonic sensors Sm and Sn. It is an argument from the first axis X in the direction from the position Kmn ((xm + xn) / 2,0)) toward the explored sound wave U (estimated sound wave E).

From this relationship, the estimated argument βmn is given by the following equation (1).
βmn = cos ^-1 (V · τmn / (xn-xm))… (1)

Here, in order to obtain the received wave reception time difference τmn from the pair of received signals Rm (t) and Rn (t), the two received signals Rm (t) and Rn (t) are used, and the received signal Rm (t) and the received signal Rm (t) are received. The cross-correlation function Cmn (τ) is calculated by the following equation (2) with the time difference τ from the signal Rn (t) as a variable.
Cmn (τ) = ∫Rm (t) ・ Rn (t + τ) dt… (2)
Then, the time difference τ at the timing when the value of the cross-correlation function Cmn (τ) exceeds the correlation threshold value Cmnth is obtained as the received wave time difference τmn1, τmn2, ...

The case where two sound sources U1 and U2 to be explored exist in the effective XY region RXY will be described with reference to FIG. A reception signal Rm (t) of the pattern shown in the column 5 (a) is obtained from the ultrasonic sensor Sm, and a reception signal Rn (t) of the pattern shown in the column 5 (b) is obtained from the ultrasonic sensor Sn. Suppose it was done. The received signals Rm (t) and Rn (t) are the received signal (tone burst signal) that arrived at the time tm1 and tun1 from the searched sound source U1 and the time tm2 and tun2 from the searched sound source U2, respectively. The received signal that arrived is included. Therefore, when the cross-correlation function Cmn (τ) is calculated by the above equation (2) using these two received signals Rm (t) and Rn (t), it is generally shown locally in the column of FIG. 5 (c). A cross-correlation function Cmn (τ) of a pattern in which a plurality (4 peaks) of parts (maximum parts) in which large values repeatedly appear can be obtained. Therefore, the correlation threshold value Cmnth shown by the alternate long and short dash line is set in the column of FIG. 5C, and the time difference τ at the timing when the value of the cross-correlation function Cmn (τ) exceeds the correlation threshold value Cmnth is the reception time difference. Let τmn1, τmn2, ... In this embodiment, since the correlation threshold value Cmnth is set to the size shown in the column of FIG. 5C, a large number of wave reception time differences τmn1, τmn2, ... Can be obtained.

From the obtained large number of received wave time differences τmn1, τmn2, ..., The estimated declination βmn1, βmn2, ... Of the large number of estimated sound sources E1, E2, ... Can be obtained by the above equation (1). That is, from the pair of received signals Rm (t) and Rn (t), a large number of estimated sound sources E1, E2, ... In addition, a large number of untrue estimated sound sources Ef (so-called false images) corresponding to non-existent explored sound sources will be detected.

As described above, although only two sound sources U1 and U2 to be explored exist in the effective XY region RXY, the estimated declinations βmn1, βmn2 ... Of a large number of estimated sound sources E1, E2, ... Can be obtained. One of the reasons is that the maximum part appears in the cross-correlation function Cmn (τ) as described above. That is, as described above, assuming that the time difference τ at the timing when the value of the cross-correlation function Cmn (τ) exceeds the correlation threshold value Cmnth is the received wave time difference τmn1, τmn2, ... This is because there are many cases where the value of Cmn (τ) exceeds this correlation threshold value Cmnth.

Another reason is that, as can be understood from FIG. 5, the calculated cross-correlation function Cmn (τ) includes the received signal Rm (t) and the received signal Rn (t) from the same explored sound source. It includes not only the cross-correlation function Cmn (τ) but also the cross-correlation function Cmn (τ) between the received signal Rm (t) from one explored sound source and the received signal Rn (t) from another explored sound source. Because it is. That is, among the received signals Rm (t) and Rn (t), the cross-correlation function Cmn (τ) is calculated for the received signals from the explored sound source U1 obtained near the time tm1 and the time tun1 to correlate with each other. In the function Cmn (τ), there is a portion (maximum portion) in which a large value appears repeatedly locally. Then, one or a plurality of received wave reception time differences τmn for the sound source U1 to be explored can be obtained from this maximum portion. Similarly, by calculating the cross-correlation function Cmn (τ) for the signals received from the explored sound source U2 obtained near the time tm2 and the time tun2, a maximum part is generated in the cross-correlation function Cmn (τ), and the cross-correlation function Cmn (τ) is covered. It is possible to obtain one or more received time difference τmn for the exploration sound source U2. However, not only these, but also by calculating the cross-correlation function Cmn (τ) between the received signal from the explored sound source U1 and the received signal from the explored sound source U2, the cross-correlation function Cmn (τ) has two. This is because a maximum portion is generated, and one or a plurality of received wave time differences τmn for a non-existent sound source can be obtained from these maximum portions. When there are P explored sound sources U1 to UP, a maximum of P ² is generated in the cross-correlation function Cmn (τ) depending on the arrangement of each explored sound source U1 and the like. , The number depends on the positional relationship between each obstacle and the ultrasonic sensors Sm and Sn.

As described above, when the estimated argument βmn1 or the like is obtained by using the cross-correlation function Cmn (τ) of the received signals Rm (t) and Rn (t), it corresponds to the genuine sound source U1 or the like to be explored. Since not only the estimated declination βmn but also the estimated declination βmn corresponding to the inauthentic estimated sound source Ef (false image) can be obtained, the declination α (α1, α2, ...) And the position U (x) of the sound source U1 to be explored and the like can be obtained. , Y) (U1 (x, y), U2 (x, y), ...) Is an obstacle to proper estimation.

Therefore, in the present embodiment, the estimation direction line generation unit 14 generates the estimation direction line Lmn, the direction line selection processing unit 15 excludes the inappropriate estimation direction line Lmn, and then the probed sound source declination acquisition unit 16 The declination α of the sound source U to be searched is estimated, and the position of the sound source U to be searched is estimated by the sound source position acquisition unit 17 to be searched.

First, the generation of the estimated direction line Lmn in the estimated direction line generation unit 14 (estimated direction line generation step ST4) will be described. As described above, the estimated declination βmn (βmn1, βmn2 ...) is a deviation from the first axis X in the direction from the starting point Kmn, which is an intermediate point between the ultrasonic sensors Sm and Sn, toward the estimated sound source E (E1 etc.). It is a horn. Therefore, in the XY plane PXY, it is possible to generate an estimated direction line Lmn (Lmn1, Lmn2 ...) Which has an estimated declination βmn (βmn1, βmn2 ...) And extends in a straight line from the starting point Kmn.

FIG. 6 shows all the obtained from each ultrasonic sensor Sm, Sn pair when the explored sound source U1 is located at the position U1 (0,7) and the explored sound source U2 is located at the U2 (3,6). The estimation direction line Lmn is described in the range of X = ± 4.0m and Y = 0 to 8.0m including the effective XY region RXY in the XY plane PXY, and the direction indicating the arrangement state of each estimation direction line Lmn. It is a line layout diagram. In the present embodiment, eight ultrasonic sensors S1 to S8 are arranged at equal intervals with an interval of 0.15 m from each other, and their average positions coincide with the origin OR00, and the four ultrasonic sensors are aligned with each other. , Are arranged symmetrically with the origin OR00 in between. Therefore, each starting point Kmn, which is an intermediate point between the ultrasonic sensors Sm and Sn, gathers within the range of X = ± 0.45 m. Further, according to FIG. 6, a large number of estimated direction lines Lmn are indicated by circles at positions U1 (0,7) and U2 (3,6) from each starting point Kmn near the origin OR00. It can be seen that it extends toward U2 or its vicinity. These estimated direction lines Lmn are genuine estimated direction lines Lmnt toward the genuine sound sources U1 and U2 to be explored. However, some of the estimated direction lines Lmn extend in a direction completely different from the positions of the sound sources U1 and U2 to be explored. These estimated direction lines Lmn are inauthentic estimated direction lines Lmnf extending toward the non-existent inauthentic explored sound source (inauthentic estimated sound source Ef).

As can be understood from FIG. 6, a large number of estimated declinations βmn acquired by the starting point declination acquisition unit 13 (starting point declination acquisition step ST3) include an inauthentic estimation direction line Lmnf toward the inauthentic estimation sound source Ef. It also includes not a few things. Therefore, when using the obtained estimated declination βmn to estimate the declination α from the origin OR00 toward the genuine probed sound source U and the position U (x, y) of the probed sound source U, as shown in FIG. In addition, there is a possibility that a problem may occur such that the false estimation sound source Ef is mistakenly included in the searched sound source U.

Therefore, in the present embodiment, the direction line selection processing unit 15 excludes an inappropriate estimated direction line Lmn (direction line selection step ST5). Specifically, first, in the division processing unit 15a (division step ST5a), the ranges of X = ± 4.0 m and Y = 0 to 8.0 m including the effective XY region RXY are predetermined in the X direction and the Y direction. It is divided by a grid spacing Ld and divided into a large number of grid regions RLxy (in this embodiment, 25,600 pieces with a grid spacing Ld of 0.05 m in the range of X = ± 4.0 m and Y = 0 to 8.0 m). (= 160 × 160) was divided into grid regions RLxy).

Next, in the counting processing unit 15b (counting step ST5b), the number Qxy of the estimated direction lines Lmn passing through the region RLxy is counted for each grid region RLxy. For example, in the enlarged lattice region RLxy shown in the upper right portion in FIG. 6, the number Qxy = 3.

In this way, the division processing unit 15a and the counting processing unit 15b divide the effective XY region RXY into a large number of grid regions RLxy, and count the number of estimation direction lines Qxy passing through each grid region RLxy. The coarse and dense state of the distribution of can be easily understood as the number Qxy associated with the lattice region RLxy.

Further, in the exclusion processing unit 15c (exclusion step ST5c), among a large number of estimation direction lines Lmn, the estimation direction lines passing through the lattice region RLxy in which the number Qxy is less than the threshold QTh are drawn by the inauthentic estimation direction line Lmnf. Presumed to exist and excluded. In this embodiment, the threshold number QTh is set to QTh = 21. For example, in FIG. 6, the lattice region RLxy enlarged and illustrated in the upper right portion is the number Qxy (Qxy = 3), which is less than the threshold number QTh. Therefore, all three estimated direction lines Lmn passing through the grid region RLxy are excluded (erased). By performing such a process for each grid region RLxy, all of the presumed inauthentic estimated direction lines Lmnf are excluded (erased) from among a large number of estimated direction lines Lmn, as shown in FIG. In addition, only the authentic estimation direction line Lmnt toward the authentic explored sound source U1 or the like can be left. Many genuine estimation direction lines Lmnt are directed to the genuine sound source U1 and the like to be explored. Therefore, if a certain estimated direction line Lmn is a genuine estimated direction line Lmnt, it is considered that the estimated direction line Lmn does not pass through the lattice region RLxy having a low number Qxy. On the contrary, it can be estimated that the estimated direction line Lmn passing through the lattice region RLxy of the number Qxy less than the threshold number QTh is not the true estimated direction line Lmnt. Thus, the authentic estimated direction line Lmnt can be easily selected.

After that, in the probed sound source declination acquisition unit 16 (explored sound source declination acquisition step ST6), the estimated direction line Lmn shown in the direction line layout diagram of FIG. 7, that is, the exclusion process in the exclusion process section 15c results in the remaining. The frequency distribution of a large number of estimated deviations βmn corresponding to the large number of authentic estimation direction lines Lmnt is examined (see FIG. 8). According to FIG. 8, it can be seen that the peak of the frequency exists at the estimated declination βmn near 63 degrees and around 90 degrees.
Therefore, the estimated declination βmn (90 degrees and 63 degrees in this embodiment) at which the frequency peaks is set as the declination α1, α2 (α1 = 90 degrees, α2 =) in the direction from the origin OR00 toward the sound sources U1 and U2 to be explored. 63 degrees).

According to FIG. 8, the peak of the frequency near 90 degrees is almost one to the true value of the declination of 90 degrees (= tan ^-1 ∞) in the direction from the origin OR00 to the sound source U1 (0,7) to be explored. I am doing it. Further, the peak of the frequency near 63 degrees almost coincides with the true value of the declination of 63.4 degrees (= tan ^-1 2) in the direction from the origin OR00 toward the sound source U2 (3, 6) to be explored.
From this, it can be confirmed from FIG. 8 that it is appropriate to set the declination α1 = 90 degrees and the declination α2 = 63 degrees in the probed sound source declination acquisition unit 16.

Further, in the apparatus 1 of the present embodiment, for each of the explored sound source U1 and the like whose declination α1 and the like are obtained by the explored sound source declination acquisition unit 16, in the explored sound source position acquisition unit 17 (explored sound source position acquisition step ST7). , The position U1 (x, y) or the like, or the distance Du1 or the like from the origin (average position) OR00 and the argument α1 or the like are obtained (see FIG. 2).
The distance Du1 is, for example, the time difference between the ultrasonic SG of the tone burst wave radiated from the transmitter TM reflected by the probed sound wave U1 and reaching the ultrasonic sensors S1 to S8 (SN) and the sound wave V of the atmosphere. Can be estimated from. The position U1 (x, y) of the sound source U1 to be explored can be calculated from the distance Du1 and the declination α1 (x = Du1 · cosα1, y = Du1 · sinα1).

After that, the position U1 (x, y) of the explored sound source U1 or the like acquired by the explored sound source position acquisition unit 17, such as the declination α1 in the direction toward the explored sound source U1 or the like obtained by the explored sound source declination acquisition unit 16. ) Etc., the distance Du1 or the like and the argument α1 or the like) are transmitted to the display unit 40 and the voice unit 50 through the interface unit 18 and the CAN bus CB outside the device 1. The display unit 40 displays, for example, the position of the acquired sound source U1 or the like, a warning display corresponding to this position, or the like toward the driver or the like. Further, the voice unit 50 emits a voice such as a position of the acquired sound source U1 to be searched, a warning sound corresponding to this position, and a warning message generated by the synthetic voice to the driver or the like.

Thus, in the sound source position estimation device 1 of the present embodiment, by performing appropriate processing on the obtained large number of estimated declinations βmn, the orientation (argument α1 etc.) of the genuine explored sound source U1 and the like and the sound source position U1 (the declination α1 etc.) and the sound source position U1 ( The distance Du1 and the like and the argument α1 and the like can be easily estimated such as x, y). These advantages are also in the sound source position estimation method used in the present embodiment, and in the present embodiment, the control is made to function as a processing unit (oscillation control unit 11 to interface unit 18) that performs various processes by the program PG1. The same applies to the unit 10 (computer).

(Deformed form)
In the above-described embodiment, among the control units 10 of the sound source position estimation device 1, an example in which the direction line selection processing unit 15 is provided with the division processing unit 15a, the counting processing unit 15b, and the exclusion processing unit 15c is shown. ..

On the other hand, in the present modification, among the control units 20 of the sound source position estimation device 21, the direction line selection processing unit 25 shown by the broken line in FIG. 1 includes the division processing unit 15a and the counting processing unit 15b similar to those in the embodiment. , The difference is that the weighting processing unit 25c and the selection processing unit 25d are provided instead of the exclusion processing unit 15c, and the others are the same. The same applies to the sound source position estimation method and the sound source position estimation program PG2 used in this modification.
Therefore, different parts will be mainly explained, and similar parts will be omitted or simplified.

As described above, among the control units 20 of the apparatus 21 of the present modification, the direction line selection processing unit 25 (direction line selection step ST15) is the same division processing unit 15a (division step ST5a) and counting processing unit as in the embodiment. It has 15b (counting step ST5b). That is, the division processing unit 15a (division step ST5a) divides the range of X = ± 4.0 m and Y = 0 to 8.0 m including the effective XY region RXY into a large number of grid regions RLxy having a grid spacing Ld (that is, Also in this modified form, the range of X = ± 4.0 m and Y = 0 to 8.0 m is divided into 25,600 grid regions RLxy with the grid spacing Ld = 0.05 m). Then, the subsequent counting processing unit 15b (counting step ST5b) also counts the number Qxy of the estimated direction lines Lmn passing through each lattice region RLxy, as in the embodiment. For example, in the enlarged lattice region RLxy shown in the upper right portion in FIG. 6, the number Qxy = 3.

However, in this modified form, in the subsequent weighting processing unit 25c (weighting step ST15c), N (N = 8 in this modified form) average positions on the XY plane PXY such as ultrasonic sensors S1 (in this modified form, average). The larger the distance Dxy (= √ (x ² + y ² )) from each lattice region RLxy (the position corresponds to the origin OR00), the more weighting is performed to change the number Qxy to the correction number CQxy of a large value. Specifically, in this modified form, the distance Dxy displayed in meters, such as the correction coefficient Cxy proportional to the distance Dxy to the number Qxy (for example, when Dxy = 1.5 m, the correction coefficient is 1.5). The correction coefficient Cxy = Dxy) equal to is multiplied to obtain the correction number CQxy (for example, CQxy = Dxy · Qxy).

The distribution of the correction number CQxy obtained by this weighting process is shown in FIG. In FIG. 9, it can be confirmed that the distribution of the grid region RLxy in which the value of the corrected number CQxy is slightly large due to the untrue estimated sound source Ef (false image) is partially confirmed. However, the value of the corrected number CQxy is particularly large in the vicinity of the genuine explored sound sources U1 and U2 and in the direction from the origin OR00 toward the explored sound sources U1 and U2 (CQxy ≧ 80, which is indicated as 80+ in FIG. 9). It can be seen that the region RLxy is distributed.

Then, in the selection processing unit 25d (selection step ST15d), among a large number of estimation direction lines Lmn, the estimation direction line passing through the lattice region RLxy where the correction number CQxy exceeds the threshold correction number CQTh is set by the authentic estimation direction line Lmnt. Presumed to exist and selected. In this modification, the threshold correction number CQTh is set to CQTh = 50. Therefore, for example, in the grid region RLxy illustrated in the upper right portion of FIG. 6, the correction number CQxy is less than the threshold correction number CQTh. The estimated direction line Lmn passing through this grid region RLxy is not selected at least because it passes through this grid region RLxy. By performing such processing for each lattice region RLxy, only those estimated to be the authentic estimation direction line Lmnt in any of the lattice regions RLxy are selected from among a large number of estimation direction lines Lmn. As a result, as shown in FIG. 10, only the authentic estimation direction line Lmnt toward the authentic explored sound source U1 or the like can be left.

After that, as in the embodiment, the probed sound source declination acquisition unit 16 selects a large number of estimated direction lines Lmn shown in the direction line layout diagram of FIG. 10, that is, by the selection process in the selection process unit 25d. The frequency distribution is examined for a large number of estimated deviations βmn corresponding to the authentic estimation direction line Lmnt. As a result, a frequency distribution in which a frequency peak exists at the estimated declination βmn near 63 degrees and 90 degrees, which is substantially the same as in FIG. 8, can be obtained (not shown).
Therefore, as in the embodiment, the estimated declination βmn (90 degrees and 63 degrees in this embodiment) at which the frequency peaks is set as the declination α1, α2 (α1 =) in the direction from the origin OR00 toward the sound sources U1 and U2 to be explored. 90 degrees, α2 = 63 degrees).

Further, as in the embodiment, the searched sound source position acquisition unit 17 obtains the position U1 (x, y) or the like, or the distance Du1 or the like from the origin OR00 and the declination α1 or the like (see FIG. 2). These are transmitted to the display unit 40 and the voice unit 50 through the interface unit 18 and the CAN bus CB outside the device 21.

Thus, even in the sound source position estimation device 21 of this modified form, the orientation (argument α1 etc.) and the sound source position of the genuine probed sound source U1 or the like can be easily determined by appropriately processing the obtained large number of estimated declination βmn. Can be estimated to. These advantages are also in the sound source position estimation method used in this modified form, and in this modified form, the control is made to function as a processing unit (oscillation control unit 11 to interface unit 18) that performs various processes by the program PG2. The same applies to the unit 20 (computer).

In the above, the present invention has been described according to the embodiment and the modified form, but the present invention is not limited to the above-described embodiment and the like, and various modifications are possible. The above-described embodiments and the like are exemplified for the purpose of explaining the present invention in an easy-to-understand manner, and are not necessarily limited to those having all the described configurations. Further, another configuration may be added to the configuration of the embodiment or the like, and a part of the configuration may be replaced with another configuration. Possible variants to the embodiment are, for example:

Since the hardware of the control units 10 and 20 in the embodiments and the like can be realized by a general computer, the sound source position estimation programs PG1 and PG2 that execute the above-mentioned various processes can be stored in the storage medium in advance. It may be distributed via a transmission line.
Further, various processes in the control units 10 and 20 have been described as software-like processes using the sound source position estimation programs PG1 and PG2 in the embodiments and the like, but some or all of them are specified by ASIC (Application Specific Integrated Circuit). It can also be replaced with hardware-like processing using an application-specific IC) or FPGA (Field Programmable Gate Array).

SG, SGT, SGR Ultrasonic S1, S2, ..., SN, Sm, Sn Number of ultrasonic sensors N (of ultrasonic sensors) m, n (indicating the order of ultrasonic sensors) Orders R1 (t), R2 (t) ), ..., RN (t), Rm (t), Rn (t) Received signal Cmn (τ) (received signal Rm (t) and Rn (t)) mutual correlation function Cmnth Correlation thresholds τmn, τmn1, τmn2, ... Received time difference U, U1, U2, ..., UP Searched sound source (genuine searched sound source)
P Number of (explored sound sources) E, E1, ..., Ef Estimated sound source Ef Inauthency Estimated sound source βmn, βmn1, βmn2 ... Estimated direction line Lmnt (toward the genuine sound source to be explored) Authentic estimated direction line Lmnf (toward the untrue estimated sound source) Inauthentic estimated direction line α, α1, ... , ΑP (From the origin to the sound source to be explored) Declination 1,21 Sound source position estimation device 10,20 Control unit (computer)
PG1, PG2 Sound source position estimation program 13 Origin argument acquisition unit 14 Estimated direction line generation unit 15, 25 Direction line selection processing unit 15a Division processing unit 15b Counting processing unit 15c Exclusion processing unit 25c Weighting processing unit 25d Selection processing unit 16 Searched Sound source declination acquisition unit (argument acquisition unit)
17 Searched sound source position acquisition unit ST3 Origin declination acquisition step ST4 Estimated direction line generation step ST5, ST15 Direction line selection step ST5a Division step ST5b Counting step ST5c Exclusion step ST15c Weighting step ST15d Selection step ST6 Exploration sound source declination acquisition step X 1st axis Y 2nd axis PXY XY plane RXY Effective XY area RLxy Lattice area Qxy Number of lines (estimated direction line passing through the lattice area) QTh Threshold number OR00 Origin (average position of M ultrasonic sensors in XY plane)

Claims (12)

Using received signals obtained from N ultrasonic sensors (N is a natural number of 3 or more) arranged at intervals on the first axis, P (P is a natural number of 1 or more) to be explored. A sound source position estimation device that estimates the position of a sound source.
Using the pair of received signals obtained from the pair of ultrasonic sensors, the estimation starts from the intermediate point between the pair of ultrasonic sensors and goes from the starting point to the estimated sound source estimated to be the probed sound source. A starting point deviation acquisition unit that acquires deviations for each combination of ultrasonic sensors, and
Of the virtual XY plane formed by the first axis and the second axis extending in the direction orthogonal to the first axis from the first axis, each of the effective XY regions where the position of the sound source to be explored can be effectively estimated. An estimated direction line generator that generates an estimated direction line extending from the above starting point to the estimated sound source by making an estimated argument.
From the estimated direction lines generated in the effective XY region, the estimated direction line estimated to be the genuine estimated direction line toward the genuine sound source to be searched is selected, or the estimated direction line toward the untrue estimated sound source is not selected. A sound source position estimation device including a direction line selection processing unit that excludes an estimation direction line estimated to be a genuine estimation direction line. The sound source position estimation device according to claim 1.
The direction line selection processing unit is
A division processing unit that divides the effective XY region into a large number of grid regions having a predetermined grid spacing, and a division processing unit.
A counting processing unit that counts the number of the estimated direction lines passing through each grid region,
A sound source position estimation having an exclusion processing unit that presumes that the estimated direction lines passing through the grid region in which the number of lines is less than the threshold line is excluded from the estimated direction lines. Device. The sound source position estimation device according to claim 1.
The direction line selection processing unit is
A division processing unit that divides the effective XY region into a large number of grid regions having a predetermined grid spacing, and a division processing unit.
A counting processing unit that counts the number of the estimated direction lines passing through each grid region,
A weighting processing unit that corrects the number of the above-mentioned number to the number of corrections by a weighting correction that becomes heavier as the distance from the average position of the N ultrasonic sensors on the XY plane to the lattice region increases.
A sound source position having a selection processing unit that estimates and selects an estimated direction line passing through the grid region in which the number of corrections exceeds the number of threshold corrections among the estimated direction lines. Estimator. The sound source position estimation device according to any one of claims 1 to 3.
The N ultrasonic sensors are arranged symmetrically on the first axis with the average position in the XY plane as the center.
In the frequency distribution of the magnitude of the estimated declination of the estimated declination line selected or excluded by the direction line selection processing unit, the value of the estimated declination at which the frequency indicates a local peak is calculated from the average position. A sound source position estimation device that also has a declination acquisition unit that sets the value of the declination toward the probed sound source. Using received signals obtained from N ultrasonic sensors (N is a natural number of 3 or more) arranged at intervals on the first axis, P (P is a natural number of 1 or more) to be explored. A method for estimating the position of a sound source that estimates the position of a sound source.
Using the pair of received signals obtained from the pair of ultrasonic sensors, the estimation starts from the intermediate point between the pair of ultrasonic sensors and goes from the starting point to the estimated sound source estimated to be the probed sound source. The starting point deviation acquisition step for acquiring the deviation angle for each combination of the above ultrasonic sensors, and
In the virtual XY plane formed by the first axis and the second axis extending in the direction orthogonal to the first axis from the first axis, the effective XY region where the position of the sound source to be explored can be effectively estimated is above. An estimated direction line generation step that generates an estimated direction line extending from each of the above starting points to the estimated sound source by making an estimated argument.
From the estimated direction lines generated in the effective XY region, the estimated direction line estimated to be the genuine estimated direction line toward the genuine sound source to be searched is selected, or the estimated direction line toward the untrue estimated sound source is not selected. A sound source position estimation method comprising a direction line selection step that excludes an estimated direction line that is presumed to be a genuine estimated direction line. The sound source position estimation method according to claim 5.
The direction line selection step is
A division step of dividing the effective XY region into a large number of grid regions having a predetermined grid spacing, and
A counting step that counts the number of the estimated direction lines passing through each grid region,
A sound source position estimation method including an exclusion step of presuming that the estimated direction lines passing through the grid region in which the number of lines is less than the threshold line are excluded from the estimated direction lines by presuming that they are the inauthentic estimated direction lines. .. The sound source position estimation method according to claim 5.
The direction line selection step is
A division step of dividing the effective XY region into a large number of grid regions having a predetermined grid spacing, and
A counting step that counts the number of the estimated direction lines passing through each grid region,
A weighting step of correcting the number of the above-mentioned number to the number of corrections by a weighting correction that becomes heavier as the distance from the average position of the N ultrasonic sensors on the XY plane to the grid region increases.
A sound source position estimation having a selection step of estimating and selecting an estimated direction line passing through the grid region in which the number of corrections exceeds the number of threshold corrections among the estimated direction lines is estimated to be the genuine estimation direction line. Method. The sound source position estimation method according to any one of claims 5 to 7.
The N ultrasonic sensors are arranged symmetrically on the first axis with the average position in the XY plane as the center.
In the frequency distribution of the magnitude of the estimated declination of the estimated declination line selected or excluded in the direction line selection step, the value of the estimated declination at which the frequency indicates a local peak is calculated from the average position. A sound source position estimation method that also includes a declination acquisition step that is the value of the declination toward the exploration sound source. Using received signals obtained from N ultrasonic sensors (N is a natural number of 3 or more) arranged at intervals on the first axis, P (P is a natural number of 1 or more) to be explored. A sound source position estimation program that estimates the position of a sound source.
Computer,
Using the pair of received signals obtained from the pair of ultrasonic sensors, the estimation starts from the intermediate point between the pair of ultrasonic sensors and goes from the starting point to the estimated sound source estimated to be the probed sound source. A starting point deviation acquisition unit that acquires deviations for each combination of ultrasonic sensors, and
Of the virtual XY plane formed by the first axis and the second axis extending in the direction orthogonal to the first axis from the first axis, each of the effective XY regions where the position of the sound source to be explored can be effectively estimated. An estimated direction line generator that generates an estimated direction line extending from the above starting point to the estimated sound source by making an estimated argument.
From the estimated direction lines generated in the effective XY region, the estimated direction line estimated to be the genuine estimated direction line toward the genuine sound source to be searched is selected, or the estimated direction line toward the untrue estimated sound source is not selected. A sound source position estimation program that functions as a direction line selection processing unit that excludes the estimation direction line that is estimated to be the authentic estimation direction line. The sound source position estimation program according to claim 9.
The computer
In functioning as the direction line selection processing unit,
A division processing unit that divides the effective XY region into a large number of grid regions having a predetermined grid spacing, and a division processing unit.
A counting processing unit that counts the number of the estimated direction lines passing through each grid region,
Of the estimated direction lines, a sound source that functions as an exclusion processing unit that presumes that the estimated direction lines passing through the grid region in which the number is less than the threshold is excluded as the inauthentic estimated direction lines. Position estimation program. The sound source position estimation program according to claim 9.
The computer
In functioning as the direction line selection processing unit,
A division processing unit that divides the effective XY region into a large number of grid regions having a predetermined grid spacing, and a division processing unit.
A counting processing unit that counts the number of the estimated direction lines passing through each grid region,
A weighting processing unit that corrects the number of the above-mentioned number to the number of corrections by a weighting correction that becomes heavier as the distance from the average position of the N ultrasonic sensors on the XY plane to the lattice region increases.
Among the estimated direction lines, the estimated direction lines passing through the grid region where the number of corrections exceeds the threshold correction number are presumed to be the genuine estimated direction lines and are selected as a selection processing unit. Sound source position estimation program. The sound source position estimation program according to any one of claims 9 to 11.
The N ultrasonic sensors are arranged symmetrically on the first axis with the average position in the XY plane as the center.
The computer
In the frequency distribution of the magnitude of the estimated declination of the estimated declination line selected or excluded by the direction line selection processing unit, the value of the estimated declination at which the frequency indicates a local peak is calculated from the average position. A sound source position estimation program that also functions as a declination acquisition unit that sets the value of the declination toward the probed sound source.

Images