JP2003014459A - Azimuth detector, azimuth detecting method and walking detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an azimuth detector, an azimuth detecting method and the like capable of performing a precise azimuth detection with a simple structure. SOLUTION: In an azimuth detection sensor 10, coordinate information showing an azimuth or inclination by an optical code shown by a two-dimensional barcode, a mark or a pit is written on the surface of a sphere 22 to be influenced by geomagnetism, and it is read by optical code reading parts 26H and 26V, whereby the azimuth of a walker is detected. Further, such an azimuth detection sensor 10 can be provided on a walking detector, and this walking detector collects components of low frequency band transmitted within the body in walking by a microphone and performs a detection of walking on the basis of this to estimate the length of step of the walker, and it further detects the change of direction of the walker by detecting the azimuth by the azimuth detection sensor 10.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an azimuth detecting device, an azimuth detecting method, and a direction detecting device using them.

[0002]

2. Description of the Related Art It is well known that geomagnetism is used to detect an azimuth.
Magnetic sensors have been used conventionally. That is, the magnetic sensor detects the magnetic field lines of the earth's magnetism. By detecting the azimuth in this way, it is possible to detect a change in the direction of a moving body such as a person or a vehicle. That is, the change in the orientation of the person or vehicle wearing the magnetic sensor or the like is detected based on the displacement of the direction detected by the magnetic sensor or the like.

[0003]

However, when the azimuth is detected by the magnetic sensor as described above, it is necessary to keep the magnetic sensor horizontal with respect to the ground plane and accurately set the angle with respect to the axis of the geomagnetism. It was This is because if the magnetic sensor has a tilt, an error will occur in the detected azimuth due to the combined angle formed by the azimuth due to the geomagnetism and the tilt of the sensor. For this reason, when using a magnetic sensor or the like, particularly when it is attempted to detect a change in the orientation of a person who has a large posture change as compared with a vehicle or the like, it is practically difficult to perform highly accurate detection. In order to solve such a problem, it is conceivable to use a gyro sensor or the like and detect the tilt of the magnetic sensor to make a correction.
In this case, not only the magnetic sensor for detecting the azimuth but also a gyro sensor, a correction circuit, etc. are required, and the mechanism becomes complicated. The present invention has been made based on such a technical problem, and a main object of the present invention is to provide an azimuth detecting device, an azimuth detecting method, and the like capable of performing highly accurate azimuth detection with a simple configuration. .

[0004]

According to the above object, in the azimuth detecting device of the present invention, an optical code indicating coordinate information is written on the surface of the azimuth magnet, and the optical code is read by the optical code reading section. , The azimuth indicated by the azimuth magnet is detected. The optical code includes a two-dimensional bar code, a symbol made up of a character string, a numerical string, etc., and the coordinate information shown by these optical codes is
By associating with the azimuth indicated by the azimuth magnet, the azimuth information can be directly obtained by reading the optical code. Further, the tilt of the azimuth detecting device can be detected by including the information of the tilt obtained from the relative displacement between the azimuth magnet and the casing when the azimuth detecting device is tilted in the optical code. In addition, by interposing a liquid having a larger specific gravity than the compass between the casing and the compass, the compass is held in a floating state on the liquid, and vibration, for example ultrasonic vibration, is applied to the casing by the vibrating section. You can also This reduces the friction between the liquid and the compass.

The present invention comprises the steps of reading an optical code indicating the coordinate information written on the surface of the azimuth magnet, and detecting the azimuth previously associated with the coordinate information based on the read optical code. It can be understood as a direction detection method characterized by the above. It is also possible to continuously perform the step of reading the optical code a plurality of times and detect the vibration based on the change in the coordinate information obtained from the read optical code. Further, the acceleration can be detected based on the detected vibration.

By the way, the present applicant came up with the above-described azimuth detecting device and azimuth detecting method in the process of developing a walk detecting device for detecting the walk of a pedestrian. there were. Conventionally, as a method of detecting the walking of a pedestrian, a so-called pedometer (registered trademark) that has a built-in vibrator and detects the vibration of the vibrator during walking
There is a method. However, in a mechanical method with a built-in vibrator, the number of steps is counted even when the vibrator is not operating, and the number of steps is counted, so that the measurement accuracy is low. On the other hand, in some cases, an acceleration sensor or a gyro sensor is incorporated instead of the vibrator, and vibration (movement) is detected from the acceleration detected by these sensors to detect walking. However, in order to improve the detection accuracy by such a method, the sensor must be surely attached to a specific place such as the waist of a pedestrian. Furthermore, in the case of a method using an acceleration sensor or a gyro sensor, the detection result differs depending on the sensor orientation, so it must be set on a pedestrian before the axial detection of the sensor takes time,
The program and circuit configuration becomes complicated. Further, in order to detect a complicated movement of a pedestrian during walking with high accuracy, a multi-axis acceleration sensor is required, which leads to a complicated structure and an increase in cost.

The pedestrian's walking detected by the above-described method is used not only for counting the number of steps like a pedometer, but also, for example, there is a method for estimating the amount of movement of the pedestrian from the detected walking. This is GPS (Global Positioning)
System: Global Positioning System) in a navigation device, the position measured based on radio waves obtained from GPS satellites is corrected based on the amount of movement of a pedestrian,
It is used when performing so-called autonomous navigation. Conventionally,
In such a case, the stride is estimated by utilizing the fact that the two parameters of the pedestrian's walking frequency (the number of steps per unit time) and the height are correlated with the stride.

Further, a technique for identifying a walking detection target based on the detected walking has been proposed. In such a case, in addition to the method described above, there is a method of detecting walking by detecting footsteps. For example, when a horse walks, the sound of a hoof touching the ground is collected with a microphone.
The individual is identified from the walking mode obtained from this sound (Japanese Patent Laid-Open No. 2000-193520).

When the walking is detected from the sound collected by the microphone as described above, the walking of each step is detected by processing the detected waveform. In this case, there are some that simply count the peaks of the waveform that occurs when walking, and by converting the time-series changes in the waveform by Fourier transform or wavelet transform, convert it into an intensity spectrum pattern of the frequency and then use that pattern. There is also one that detects walking by analyzing. However, in an environment in which walking is actually detected, there is noise that is mixed in from outside the pedestrian, such as noise that occurs on the vehicle side when a pedestrian is in a vehicle. Therefore, no matter which method is used to detect walking, these noises greatly affect the detection accuracy (analysis accuracy).

Further, when the movement amount is estimated from the walk of the pedestrian, the step length is estimated from the pace and height of the pedestrian.
In the conventional method, as described above, the detection accuracy itself of walking cannot be said to be sufficiently high, and furthermore, a pedestrian walked in a pattern other than normal walking, for example, with a wide stride and stride. Since it is naturally difficult to discriminate in the case of running, running up and down stairs, etc., it is also difficult to accurately detect the stride, that is, to accurately estimate the amount of movement.

As a walking detection device capable of solving such a conventional problem, the applicant of the present invention collects a vibration generated when a pedestrian walks and converts it into an electric signal,
We propose a configuration that estimates the amount of movement of a pedestrian by analyzing the change in the signal corresponding to a predetermined frequency or lower based on this electrical signal and detecting the walking of the pedestrian. Then, in the walking detection device having such a configuration, the optical code indicating the azimuth written on the surface of the azimuth magnet is read,
By detecting the azimuth previously associated with this optical code, the change in the moving azimuth of the pedestrian is detected, and based on the detected amount of movement and the change in the moving azimuth,
It is provided with a configuration for detecting the position of a pedestrian.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described below in detail based on the embodiments shown in the accompanying drawings. FIG. 1 is a diagram showing the configuration of the azimuth detecting apparatus according to the present embodiment. As shown in FIG. 1, the azimuth detecting sensor (azimuth detecting device, azimuth detecting means) 10 includes a sensor main body 20 and a controller 30. The sensor main body 20 includes a sphere 22 inside a housing 21, and a casing 23 that houses the sphere 22 in a rotatable state.

The spherical body 22 has a built-in azimuth magnet having a directivity with respect to the earth's magnetism by being magnetized in a spherical shape, and an optical code indicating the coordinate information of the spherical body 22 when it is rotated over the entire surface thereof. Many are written. Here, the optical code written on the surface of the sphere 22 is a two-dimensional bar code as shown in FIG. 2 (a), a symbol such as a character string or numeral string as shown in FIG. 2 (b), and the like. Other than that, a recording method similar to the method used for pits used in optical discs and the like are conceivable, and the coordinate information indicated by these optical codes is indicated by the azimuth detection sensor 10 based on preset table information. The information can be directly converted into the information of the azimuth and tilt angle. This sphere 22
A sphere 2 is formed by incorporating a weight (not shown) in the bottom portion 22a.
It is formed so as to have a higher specific gravity than the other portions of 2.

The casing 23 is made of a transparent material, and its inner diameter is set to be larger than the outer diameter of the spherical body 22 by a certain dimension or more. And this casing 23
The liquid 24 is injected into the gap between the sphere 22 and the sphere 22. Here, the liquid 24 is stored in the air chamber 25 in the upper part of the casing 23.
Is formed so that the sphere 22 is formed, and the specific gravity thereof is selected to be larger than that of the spherical body 22. As a result, the sphere 22 moves from the liquid surface of the liquid 24 to the top 2 thereof.
2b is projected without contact with the casing 23, and the built-in azimuth magnet has directivity due to the influence of the earth's magnetism, and has a directivity regardless of the movement (direction) of the housing 21 side. To face.

Optical code reading sections (code reading means) 26H, 26 using laser light or the like are provided on the inner peripheral surface of the casing 21.
V is provided. These optical code readers 26
The H and 26V irradiate the surface of the sphere 22 with a laser beam or the like through the transparent casing 23 and detect the reflected light to read the optical code written on the surface of the sphere 22. These optical code readers 26H, 26
When the housing 21 is in a horizontal state, the V irradiates the spherical body 22 with laser light or the like from the horizontal direction and the vertical direction. Here, one optical code reading unit 26
H reads the coordinate information from the optical code written on the surface of the sphere 22 to obtain the direction, and the other optical code reading unit 26V obtains the tilt angle from the optical code written on the surface of the sphere 22. To read the coordinate information.

By the way, the casing 23 is supported by the casing 21 via an ultrasonic transducer (vibrating section) 27 with respect to the casing 21. The ultrasonic vibrator 27 applies ultrasonic vibration of a predetermined frequency to the casing 23, thereby reducing static friction acting between the liquid 24 and the sphere 22 in the casing 23, and moving the sphere 22. The purpose is to improve.

On the other hand, the controller 30 comprises an optical code detecting section 31, a detection control section (azimuth detecting section, inclination detecting section, azimuth change detecting means) 32, and a result output section 33. As shown in FIG. 3, the optical code detector 31 detects the optical code read by the optical code readers 26H and 26V (step S101). The detection control unit 32 acquires the information of the azimuth / tilt angle corresponding to the coordinate information indicated by the read optical code based on the table information stored in advance (step S102). Then, the result output unit 33 outputs the information on the azimuth / tilt angle converted by the detection control unit 32 to the outside, for example, the walking detection device 36 described later (step S10).
3).

In the sensor body 20 having such a structure, even if the orientation of the casing 23 integrally including the optical code reading portions 26H and 26V is changed, the spherical body 22 has an axis connecting the bottom portion 22a and the top portion 22b due to the earth magnetism. The vertical state is always maintained in a state where it is oriented in a predetermined azimuth, which causes relative displacement between the optical code reading units 26H and 26V and the sphere 22. Then, the optical code reading units 26H and 26V read the optical code written on the surface of the sphere 22 in a state where the relative displacement with the sphere 22 occurs, so that the controller 30 causes the direction of the sensor main body 20 to change. Information on the tilt angle is obtained.

In the present embodiment, the walking detection device 36 as shown in FIG. 4 is provided with the azimuth detection sensor 10 described above. FIG. 4 is a diagram for explaining the basic configuration of the walking detection device 36 in the present embodiment. As shown in FIG. 4, the walking detection device 36 is a microphone (conversion means) that collects ambient sounds and converts them into electrical signals.
40, a low-pass filter 41 that passes only signals having a frequency equal to or lower than a predetermined frequency, a converter 42 that A / D-converts the sound that has been collected by the microphone 40 and that has passed through the low-pass filter 41 to convert it into a digital waveform, As described below,
An analysis unit (detection unit) 43 that detects walking by analyzing the waveform converted by the converter 42, a database 44 that stores data used for analysis by the analysis unit 43, and data of detection results by the analysis unit 43 The detection controller 3 shown in FIG.
2 and the azimuth detecting sensor 10
And are equipped with.

The microphone 40 collects vibrations (including sound) in all frequency bands that can be collected by the microphone 40. Here, by collecting ambient sound with the microphone 40, vibrations that are generated during walking and transmitted through the body of a pedestrian (wearer) and sounds transmitted through the air are collected. Usually, in a microphone of a mobile phone terminal, etc., outside the audible range (for example, 20 Hz or less and 2000
The signal in the frequency band of 0 Hz or more) is cut by the bandpass filter, but in the present embodiment, the signal output from the microphone 40 is only the signal having a frequency determined by the lowpass filter 41, for example, 200 Hz or less. Is passed through and is separated from a component in a frequency band higher than 200 Hz, a so-called audio signal. Then, the analysis unit 43 analyzes the signal of the component having the determined frequency or less.

The analysis unit 43, based on the signals converted by the converter 42, a spectrogram processing unit 51 for generating a spectrogram, a walking detection unit 52 for detecting walking of a pedestrian, and a stride estimation unit 53 for estimating stride length. And a position estimation unit (position detection unit, movement amount estimation unit, position detection unit) 5 that estimates the position of the pedestrian based on the stride obtained by the stride estimation unit 53 and the azimuth detected by the azimuth detection sensor 10.
4 and a walking mode detection unit 55 that detects the walking mode of a pedestrian
And a walking mode determination unit 56 that determines the walking mode of the pedestrian based on the detection result of the walking mode detection unit 55.

Next, in the analysis unit 43, the microphone 40
The basic concept for performing analysis based on the sounds collected in Section 2. That is, the vibration (sound) when a pedestrian walks and the left foot and the right foot are alternately landed is collected by the microphone 40, and this vibration is A / D converted by the converter 42 into a digital signal. When the signal (voltage) is wavelet-transformed, a spectrogram that is an intensity spectrum pattern of frequency as shown in FIG. 5 is obtained.

Here, FIG. 5 shows a spectrogram when the same pedestrian walks normally. The term "normal" as used herein refers to a case in which a pedestrian walks on a flat ground without being fast or slow, and without being particularly aware of the pace. As shown in FIG. 5, when a person walks, the impact that is caused by the landing of the foot and the movement of the foot over the ground, and the conduction to the joints, skeletons, and muscles of the foot such as the hips, hips, knees, ankles, toes, etc. And audible area (about 20 to 2000
A characteristic time / frequency / intensity pattern is generated in the vicinity of a low frequency band of about 100 Hz or less, more specifically, 30 to 40 Hz or less, which is close to the lower limit of 0 Hz). In the present embodiment, walking detection is performed by analyzing the pattern of this low frequency band component.

[0024] The walking motion of a human is the heel landing (Heel-Strik
e) to ground movement until the toe is released (Toe-Off), and Toe
-It can be divided into two phases of kick-out motion from Off to heel landing. Then, the landing of the bottom of the foot during ground movement starts from the heel, and the lateral part of the foot (external longitudinal arch)
Then, the direction is changed from the base of the little finger (fifth finger metatarsal floor) to the base of the thumb (first finger metatarsal bone) via the route. The ground impact caused by this ground motion is absorbed by each joint, muscle, and fat layer, but a part of the low-frequency component resonates with these components and conducts through the human body, as shown in the spectrogram of FIG. It emits vibrations in the low frequency band with complicated time, frequency and intensity patterns.

Using the example of the pedestrian shown in FIG. 5, the low frequency band components generated during walking will be examined in detail. Figure 5
In (a), a clear peak appears, and each peak corresponds to each step of walking. When the spectrogram shown in FIG. 5 is examined in more detail, the range indicated by (a) in the horizontal axis direction is the ground motion time of the right foot, the range indicated by (c) is the ground motion time of the left foot, and the range indicated by (d) is , The gap time when the grounding foot changes from the right foot to the left foot,
The range indicated by (e) is the gap time when the grounding foot changes from the left foot to the right foot. These (a), (c), (d),
The walking mode of the pedestrian can be characterized by the temporal characteristic consisting of the time length of (e). Here, in the spectrogram, the darker the color, the higher the frequency intensity, and the lighter the color, the lower the frequency intensity.

FIG. 6 shows spectrograms when the same pedestrian walks in various walking modes. Figure 6
In (a), when stepping on the spot, (b) walks late,
(c) is normal walking, (d) is fast walking, (e) is jogging,
(f) is a spectrogram detected when going up the stairs, and (g) is a spectrogram detected when going down the stairs. Comparing the spectrograms of these gait patterns, the gait cycles in FIG.
In (b), (c), and (d), the ground movement time of (a) and (c) is different (the faster the walking, the shorter the walking). In addition, FIG.
In the jogging of (e), the range of high frequency intensity extends to the high frequency region, and the peak frequency (a) of the signal is also higher than other cases. In the ascending stairs of FIG. 6 (f), the frequency intensities (f) and (f) when the foot is grounded are weak. This is because the impact at the time of touchdown is weaker than when walking on a flat surface because the foot is lifted and touches the stairs. In addition, when going up the stairs, the toes are mainly grounded, and the frequency strength at the time of grounding the heel of (f) is particularly weak. Further, comparing the ascending and descending steps of the stairs in FIGS. 6F and 6G, the band extends to a significantly higher frequency band particularly in the case of descending the stairs. This means that the energy spectrum is also generated in the high frequency band according to the impact on the foot. On the other hand, when descending the stairs in FIG. 6 (g), the toes come in contact with the ground before the heels, so that the frequency strength at the time of contacting the toes in (K) is strong.

By the way, as can be seen from the spectrogram of FIG. 5, a band-shaped pattern extending over a wide band from a low frequency outside the audible range to the audible range during normal walking is about 1 per step in association with the series of ground movements described above. Continue for about a second. Further, in the band-shaped time-frequency-intensity pattern, intensity spectra in different frequency regions corresponding to the ground movement of each part of the sole can be seen.

FIG. 7 shows the processing in the walking detection section 52. In this processing, the ground movement continuation time for each step of the sole obtained from the time, frequency and intensity in this spectrogram ((a) in FIG. 5). , (C)), the number of steps per unit time is calculated. At this time, as shown in FIG. 5, the frequency intensity continues for a certain time or more depending on the landing time of the left and right soles, and the spectrogram of time, frequency, and intensity draws a pattern peculiar to walking ( (These data are stored in the database 44 in advance) to detect the number of steps per unit time and the walking cycle, and from these characteristics of time, frequency, and intensity,
For example, impacts other than walking, such as a pedestrian's hand hitting the body, and extraneous noise caused by a vehicle or the like are distinguished from sounds generated by walking.

First, the signal w (t) obtained by A / D converting the signal from the microphone 40 by the converter 42 is used.
From the above, the low-pass filter 41 removes a signal of a predetermined frequency or higher, that is, a normal audio signal, and obtains a signal w ′ (t) (step S201). Then, the analyzing unit 43 converts this signal w ′ (t) into a functional expression p of the intensity of time and frequency.
It is converted to (t, f) (step S202). Subsequently, the functional expression p (t, f) is integrated with respect to the preset frequency band B, and the pedestrian status is determined from the relationship with the frequency intensity α. If the integrated value L _B (t) is larger than the frequency intensity α, the pedestrian status: status (t) is set to “landing state”, and if the integrated value L _B (t) is equal to or less than the frequency intensity α, walking The status of the person is set to a state other than the "normal state", that is, the "landing state" (step S203).

After this, the following step S204 is executed each time a new signal is input at every minute time. In this step S204, the signal at the time t at that time is not a signal not caused by walking, that is, a signal having characteristic points (pattern, intensity distribution) peculiar to walking as shown in FIG. Then, it is determined whether or not the "landing state" is determined (step S204). As a result, if this condition is satisfied, it is judged as an erroneous determination,
The pedestrian status and the "landing state" are changed (step S205). On the other hand, when the condition of step S204 is not satisfied, the pedestrian status is not changed. In this way, each time a signal is input, the status pedestrian status (t) at that time is determined, and the status "status (t-1)" immediately before is changed from "normal status" to "landing status". If it has changed, the number of times of landing is counted once (step S206). If the number of times of landing per unit time is calculated from the count value of the number of times of landing, the number of steps s per unit time can be obtained, and if the unit time is divided by the number of steps s, the walking cycle can be obtained. (Step S207).

Next, in the walking mode detecting section 55, as shown in FIG.
A process of recognizing the walking mode of the pedestrian from the sound collected by the microphone 40 based on the characteristics of various walking modes as shown in FIG. In this case, the data of the characteristic points of various walking modes are stored in the database 44 in advance.
As shown in FIG. 8, in the analysis unit 43, step S in FIG.
The detected characteristic pattern W of walking is obtained from the functional expression p (t, f) of the intensity of time and frequency obtained in the same manner as 201 to S202 (step S301). Then, with respect to the obtained characteristic pattern W, the database 44 is referred to, and various characteristic patterns W stored in the database 44 are referred to.
The pattern recognition processing is performed for _k , for example, the slow walking characteristic pattern W ₁ , the normal walking characteristic pattern W ₂ , the fast walking characteristic pattern W ₃ ,. For this, for example, the characteristic pattern W
_k and each characteristic pattern W _k stored in the database 44
Mahalanobis distance is calculated to obtain the difference d _k from the characteristic pattern W (step S302). Then, the smallest difference d _k from each feature pattern W _k is searched (step S303), and the walking pattern of the pedestrian is estimated from the feature pattern W _k having the smallest difference d _k (step S304). .

In order to estimate the step length by the step length estimating section 53, processing based on the detection results of the walking detecting section 52 and the walking mode detecting section 55 is performed. As one method of the processing here, there is a method in which the walking detection unit 52 detects the walking frequency and estimates the stride from the walking frequency and the height of the pedestrian. It is generally known that the walking frequency (step) and the height of the pedestrian are correlated with the stride, as shown in FIG. In FIG. 9, the walking frequency and height,
It is a plot of the stride relationship, and as a whole, it is clear that they have a correlation.

In this case, the database 44 stores the data of the relational expression (coefficient) of the walking frequency, the height of the pedestrian, and the stride obtained from the correlation shown in FIG. Then, the height (and sex) of the pedestrian is input to the setting memory 57 before detection of walking. To estimate the stride, as shown in step S401 of FIG. 10, the walking cycle detected by the walking detection unit 52 and the setting memory 57 are set.
Based on the height of the pedestrian set to, the gait model M is used to predict the step length l. The walking model M is represented by, for example, the following relational expression. Stride length l (cm) = x1 x walking cycle (steps / min) + x2 x height (c
m) + C In the above relational expression, x1, x2, and C are coefficients and initial values that are stored in the database 44 in advance and set for each walking mode.

Then, the position estimating unit 54 integrates the step length l thus obtained and the number of steps s per unit time detected by the walking detecting unit 52 to obtain the moving distance per unit time. Can be estimated (step S402). At this time, in the direction detection sensor 10,
By continuously detecting the azimuth, it is possible to detect the azimuth change when the pedestrian walks. Then, based on the estimated moving distance and the detected azimuth change, the position estimating unit 54 can estimate the moving route of the pedestrian. In other words, if the position of the pedestrian at the start of detection by the walking detection device 36 is known, it is possible to estimate the position of the walking pedestrian.

Furthermore, in step S401, an example of the relational expression of the walking model M is given, but the present invention is not limited to this. For example, the relational expression of the walking model M can be determined based on the ratio of the energy when the heel lands and the energy when the toe is kicked out. That shown in Figure 11, it represents the change in energy when walking, wherein the energy E ₁ kicking the energy when kicking the toe is the peak value of the walk signal. Further, the heel landing energy E ₂ when landing on the heel is the maximum value immediately before the peak value corresponding to the kicking energy E ₁ . Then, the heel landing energy E ₂ divided by the kicking energy E ₁ is the landing / kicking energy ratio. Based on the landing / kicking-out energy ratio obtained in this way, the walking cycle detected by the walking detection unit 52, and the height of the pedestrian set in the setting memory 57, the stride 1 is calculated using the walking model M ′. Can be predicted. The walking model M ′ is represented by, for example, the following relational expression. Stride length l (cm) = x1 x walking cycle (steps / min) + x2 x height (c
m) + x3 × landing / kicking-out energy ratio In the above relational expression, x1, x2, and x3 are coefficients for each walking mode stored in the database 44 in advance.

As another method of estimating the step length in the step length estimating section 53, there is a method of utilizing a change in frequency intensity caused by an impact when the heel or the toe part lands. In the walking mode of humans, walking with fast walking increases both the walking frequency and the stride, but when walking consciously with a large stride, the walking frequency only changes and the walking frequency itself does not change. Further, when moving a long distance, the stride becomes a substantially constant steady value determined by the height and the gait, but the stride easily changes depending on the short moving distance or the mood at that time. In order to avoid such an influence, as shown in FIG. 5, the frequency intensity at the time of normal walking is stored in advance in the database 44 as a reference value, and the heel or toe is landed by comparison with this reference value. The increase / decrease in the amount of impact when performing is detected, and the stride length is corrected.

FIG. 12 shows a specific flow of the above processing. First, step S401 in FIG.
Similarly, the step length 1 is predicted using the walking model M or M ′ based on the walking cycle detected by the walking detection unit 52 and the height of the pedestrian set in the setting memory 57 (step S501). Then, steps S201 to S20 of FIG.
The relational expression p of the intensity of time and frequency obtained in the same manner as in 2.
(t, f) and 'integrated with respect to impact L _B of the landing' frequency bands B that can be characteristically extracting the impact of the landing obtaining (step S502). Here, what is obtained by integrating the relational expression p (t, f) is the occupation rate (area) of the spectrogram signal in the frequency band B ′. Database 44 in advance
The stride correction value Δl is calculated from the impact L _B 'based on the table or the relational expression stored in (step S
503). Then, the travel distance (1 × s + Δl) is obtained by using the stride correction value Δl (step S504). In this way, the stride can be corrected from the frequency intensity of the spectrogram.

Further, as another method of estimating the step length in the step length estimating section 53, there is a method of utilizing the frequency intensity at the time of kicking the toe part and at the time of landing the toe part. That is, even when walking in the same step, the stride varies depending on the kicking strength of the toes. Therefore, the reference value of the frequency intensity of the pedestrian stored in the database 44 in advance or the walking until now according to the intensity of the frequency band around 10 to 16 Hz, which occurs according to the landing and kicking of the toe. The stride is corrected by comparing the history data.

FIG. 13 shows the flow of such processing, and similar to steps S501 to S503 of FIG.
The walking cycle detected by the walking detector 52 and the setting memory 5
After predicting the stride 1 using the walking model M or M ′ based on the height of the pedestrian set to 7 (step S60
1), the relational expression p (t, f) of the intensity of time and frequency is integrated with respect to the frequency band B ′ capable of characteristically extracting the impact of landing to obtain the impact L _B ′ of landing (step S602). . Further, based on the table or the relational expression stored in advance in the database 44, the stride correction value Δl is calculated from the impact L _B '(step S603). After that, in the analysis unit 43, a predetermined low frequency band f (for example, 0 to
40 Hz), the characteristic of the relational expression p (t, f) is extracted (step S604), and the stride is calculated from the characteristic pattern data of the frequency intensity, the reference value of the pedestrian, and the walking history data stored in the database 44. The correction value h of is acquired (step S605). After that, the correction value h can be used to obtain the moving distance (1 × s + Δl + h).
(Step S606).

According to the walking detection device 36, the microphone 40 collects the low-frequency band components transmitted through the body during walking, and the walking is detected based on the collected components to detect the stride of the pedestrian. The direction change of the pedestrian can be detected by estimating the direction and detecting the direction by the direction detection sensor 10. Thereby, the pedestrian's walking path (from the detection start position such as the start point) can be detected and the pedestrian's position can be estimated.

At this time, in the azimuth detecting sensor 10, the coordinate information indicating the azimuth and the inclination angle is written on the surface of the sphere 22 affected by the geomagnetism by an optical code indicated by a two-dimensional bar code, a symbol, a pit or the like. Therefore, by reading this, it is possible to directly detect the direction of the pedestrian, and it is possible to easily eliminate the influence of the inclination. Moreover, such an azimuth detecting sensor 10 can detect the azimuth and the tilt angle at the same time by reading the relative displacement of the sphere 22 with respect to the housing 21 by the optical code reading units 26H and 26V, and the structure is very simple. And can be easily downsized. Furthermore, the direction detection sensor 10
Since the ultrasonic vibrator 27 is provided in and the ultrasonic vibration is applied, the movement of the sphere 22 can be improved.
It is possible to reduce the diameter of the sphere 22 itself, which can also contribute to downsizing of the azimuth detection sensor 10 itself.

In the above embodiment, the azimuth detecting sensor 10 is provided with the optical code reading units 26H and 26V, but the coordinate information indicated by the optical code is associated with the azimuth information in consideration of the inclination in advance. If so, it is possible to omit the optical code reading unit 26V. Further, in the above-described embodiment, in order to hold the spherical body 22 in a floating state in the casing 23, the liquid 24 is injected into the casing 23 so as to form the air chamber 25 in the upper portion. At this time, it is important to prevent the spherical body 22 from coming into contact with the casing 23 by setting the specific gravity of the liquid 24 and the spherical body 22 and the volume of the air chamber 25. Also, instead of such a configuration,
As shown in FIG. 14, the casing 23 may be filled with two kinds of liquids 24A and 24B having different specific gravities and not being mixed or not mixed with each other. At this time, by making the specific gravity of the liquid 24A larger than that of the sphere 22 and the specific gravity of the liquid 24B smaller than that of the sphere 22, the sphere 22 can be held in a floating state.

By the way, in the above embodiment, the pedestrian's direction change based on the direction detected by the direction detection sensor 10,
Furthermore, although the configuration is such that movement is detected, the present invention is not limited to this and can be applied to other applications. For example,
By detecting the changes over time in the azimuth and tilt angles in the optical code reading units 26H and 26V, the azimuth detection sensor 10 can detect vibration. That is, as shown in FIG. 15, the optical code reading unit 26
The optical code is continuously read at H and 26V,
The optical code detector 31 of the controller 30 detects each read optical code (step S
701). The detection control unit 32 converts the optical code read by the optical code reading unit 26H into corresponding azimuth / tilt angle information based on table information stored in advance. Then, the displacement of the horizontal direction component (X, Y direction) is detected from the optical code detected a plurality of times. Similarly, the optical code read by the optical code reading unit 26V is converted into information on the corresponding azimuth and tilt angle. Then, from the optical code detected multiple times, the vertical component
The displacement in the (Y, Z directions) is detected (steps S702 to S702).
703). Then, from the displacements of these horizontal and vertical components, X, Y, Z
It detects vibration in three directions and outputs it to the outside.
(Step S704)

Further, the azimuth detecting sensor 10 can detect acceleration. That is, in the azimuth detecting sensor 10 having the configuration as shown in the above-described embodiment, the sphere 22 tilts when acceleration is applied, and this is detected. For this purpose, as shown in FIG. 16, the optical code reading units 26H and 26V continuously read the optical codes, and the optical code detecting unit 31 of the controller 30 detects the read optical codes (steps). S801). The detection control unit 32 converts the optical code read by the optical code reading unit 26H into corresponding azimuth / tilt angle information based on table information stored in advance. Then, the displacement of the horizontal direction component (X, Y direction) is detected from the optical code detected a plurality of times. Similarly, the optical code read by the optical code reading unit 26V is converted into information on the corresponding azimuth and tilt angle. Then, the displacement of the vertical direction component (Y, Z direction) is detected from the optical code detected a plurality of times (step S).
802-S803). Then, from the displacements of these horizontal and vertical components, in the azimuth detection sensor 10,
Vibrations in three directions of X, Y and Z are detected (step S80).
4). Furthermore, in order to remove high-frequency components that can be judged to be an instantaneous disturbance from the displacements of these horizontal and vertical components, displacements below a predetermined threshold value are filtered by a low-pass filter (not shown) or the like (step S805). Thereafter, the acceleration is calculated from the displacements of these horizontal and vertical components and the elapsed time, and this is output to the outside (step S806). Other than this, the configurations described in the above embodiments can be selected or changed to other configurations without departing from the gist of the present invention.

[0045]

As described above, according to the azimuth detecting device and the azimuth detecting method of the present invention, it is possible to detect the azimuth with high accuracy with a simple structure.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of an azimuth detection sensor in the present embodiment.

FIG. 2 is a diagram showing an example of an optical code written on the surface of a sphere.

FIG. 3 is a diagram showing a flow of detecting an azimuth / tilt angle by reading an optical code.

FIG. 4 is a diagram showing a configuration of a walking detection device.

FIG. 5 is an example of a spectrogram obtained from a signal continuously obtained when a pedestrian walks.

FIG. 6 is a diagram showing spectrograms in various walking modes of the same pedestrian.

FIG. 7 is a diagram showing a flow of processing when detecting the number of steps.

FIG. 8 is a diagram showing a flow of processing when estimating a walking mode.

FIG. 9 is a data distribution diagram showing the correlation among walking frequency, step length, and height.

FIG. 10 is a diagram showing a flow of processing when estimating a moving distance.

FIG. 11 is a diagram showing energy changes during walking.

FIG. 12 is a diagram showing the flow of another process when estimating the movement distance.

FIG. 13 is a diagram showing the flow of still another process when estimating the movement distance.

FIG. 14 is a diagram showing another example of the orientation detection sensor.

FIG. 15 is a diagram showing a flow when detecting vibration.

FIG. 16 is a diagram showing a flow when detecting acceleration.

[Explanation of symbols]

10 ... Direction detection sensor (direction detection device, direction detection means),
20 ... Sensor main body, 22 ... Sphere (direction magnet), 23 ... Casing, 24, 24A ... Liquid, 26H, 26V ... Optical code reading section (code reading means), 27 ... Ultrasonic transducer
(Vibration unit), 30 ... Controller, 32 ... Detection control unit (azimuth detection unit, tilt detection unit, azimuth change detection means), 36 ... Walking detection device, 40 ... Microphone (conversion means), 41 ...
Low-pass filter, 43 ... Analysis unit (detection unit), 53 ... Stride estimation unit, 54 ... Position estimation unit (position detection unit, movement amount estimation unit, position detection unit)

Claims (7)

[Claims] 1. A bearing magnet having directivity to the earth's magnetism and having an optical code indicating coordinate information on its surface, a casing rotatably holding the bearing magnet, and a bearing magnet described in the bearing magnet. An azimuth detecting device comprising: an optical code reading unit that reads the optical code; and an azimuth detecting unit that detects the azimuth indicated by the azimuth magnet based on the read optical code. 2. The azimuth detecting device according to claim 1, further comprising a tilt detecting unit that detects a tilt of the azimuth detecting device based on the optical code read by the optical code reading unit. 3. A liquid, which is interposed between the casing and the compass and has a larger specific gravity than that of the compass, and a vibrating section for vibrating the casing. 1. The direction detection device according to 1. 4. A microphone that collects vibrations generated when a wearer of the azimuth detecting device walks and converts the vibrations into an electric signal, and an electric signal transferred from the microphone,
A detector that detects a wearer's walking by analyzing a change in a signal corresponding to a predetermined frequency or less, and a detector that detects the wearer's walking movement amount, a walking movement amount detected by the detection unit, and the azimuth detection unit. The azimuth detecting device according to claim 1, further comprising: a position detector that detects the position of the wearer based on a change in the azimuth obtained by continuously detecting the azimuth of the wearer. 5. A azimuth detecting method in an azimuth detecting device in which an azimuth magnet having directivity to the earth's magnetism is rotatably held, and a step of reading an optical code indicating coordinate information written on the surface of the azimuth magnet. And a step of detecting an azimuth previously associated with the coordinate information based on the read optical code, the azimuth detecting method. 6. The method according to claim 5, further comprising the step of successively performing the step of reading the optical code a plurality of times, and detecting vibration based on a change in coordinate information obtained from the read optical code. The direction detection method described. 7. A conversion unit that collects vibrations generated when a pedestrian walks and converts the vibrations into an electric signal, and a change of a signal corresponding to a predetermined frequency or less based on the electric signal transferred from the conversion unit. A moving amount estimating means for analyzing the pedestrian's walking and estimating the moving amount of the pedestrian, an azimuth detecting means for rotatably holding an azimuth magnet having directivity to the earth's magnetism, and the azimuth magnet. The code reading means for reading the optical code indicating the azimuth written on the surface of the pedestal, and the change in the moving azimuth of the pedestrian by detecting the azimuth previously associated with the optical code based on the read optical code. A walking detection device comprising: an azimuth change detection unit that detects the position; and a position detection unit that detects the position of a pedestrian based on the change in the movement amount and the movement direction.