JP5695436B2 - Portable terminal, program and method for determining direction of travel of pedestrian using acceleration data during swing phase - Google Patents

Description

  The present invention relates to a technique for determining a traveling direction of a pedestrian using an acceleration sensor and a geomagnetic sensor. In particular, the present invention relates to an autonomous navigation technique for deriving a traveling direction in real time.

  Conventionally, there is an autonomous navigation technique that derives a traveling direction and a current position in real time using an acceleration sensor and a direction sensor. Autonomous navigation technology is combined with GPS (Global Positioning System) technology and is mainly used for a car navigation system. The car navigation system displays an accurate traveling direction and current position and driving route guidance to a destination on a display unit for a driver of a car.

  The car navigation system corrects the current position information measured by the GPS by an autonomous navigation technique such as a vehicle speed pulse or a gyro. Further, the road map information is read out as necessary, and the traveling direction and the current position are corrected so that the current travel route coincides with the road (refer to map matching technology based on a projection method, for example, Patent Document 1). As a result, it is possible to prevent the current position from being a position not on the road due to a sensor error.

  On the other hand, there is a system in which such navigation technology is applied to a portable terminal possessed by a pedestrian. Specifically, a cumulative current position from the starting point is derived using the detected “number of steps” of the pedestrian and the “step length” of the pedestrian (see, for example, Patent Document 2). When the autonomous navigation technology is applied to a pedestrian, acceleration components other than horizontal movement are also detected. Thus, the measured distance is derived from the number of steps and the step length, rather than simply integrating the output of the acceleration sensor.

The “number of steps” is detected by taking the acceleration for each axis detected by the acceleration sensor in the mobile terminal as the square root of the sum of squares (√ (x ² + y ² + z ² )) and taking the peak-to-peak as one step (for example, (See Patent Document 3). The “step length” is preset by the user or estimated from the height of the user. Alternatively, according to another technique, there is a technique of calibrating the stride by causing a pedestrian to walk a specified distance (see, for example, Non-Patent Document 1).

  The “traveling direction” is detected by the “direction sensor”. As the direction sensor, a geomagnetic sensor is generally used. There is a technique for determining the traveling direction of a pedestrian using the relationship between the vertical acceleration and the traveling direction acceleration after determining the vertical direction to derive the terminal posture (see, for example, Patent Document 4). There is also a technique for determining the traveling direction from the attitude of the terminal at a specific time (see, for example, Patent Document 5). Furthermore, there is a technique for determining the traveling direction based on the horizontal acceleration distribution (see, for example, Patent Documents 6 and 7). Further, there is a technique for determining the traveling direction from the characteristics of the pedestrian's arm swing (see, for example, Patent Document 8).

  As a past invention by the same applicant as the present application, a normal vector (direction reference vector) for an acceleration plane based on arm swing motion and a normal vector (direction reference vector) for an azimuth reference plane based on a gravity vector and a geomagnetic vector. ) To calculate the azimuth angle in the traveling direction (see Patent Document 9).

JP-A-5-061408 Japanese Patent Laid-Open No. 9-089584 Japanese Patent Laying-Open No. 2005-038018 JP 2008-039619 A WO2006 / 104140 Japanese Patent No. 4126388 JP 2008-116315 A JP 2010-271167 A JP 2010-008103 A

"Nike + iPod User's Guide", page 27, "online", [Search February 3, 2009], Internet <URL: http://manuals.info.apple.com/en/nikeipod_users_guide.pdf>

  According to the technique described in Patent Document 4, a direction in which acceleration exceeding a predetermined threshold is detected is set as a traveling direction candidate. However, when the acceleration distribution has a width rather than a straight line, it is necessary to evaluate a large number of candidates in the traveling direction, and the amount of calculation becomes enormous.

  Further, according to the techniques described in Patent Documents 5, 6, and 7, assuming that the posture of the mobile terminal worn by the pedestrian is approximately constant, the horizontal direction detected during walking It utilizes the fact that acceleration is distributed in the direction of travel. However, in practice, the posture of the mobile terminal carried by the pedestrian is not always constant. For this reason, the acceleration in the horizontal direction is often not distributed in the traveling direction.

  Furthermore, according to the technique described in Patent Document 8, the pedestrian determines the traveling direction by swinging the arm holding the mobile terminal, but it is assumed that the mobile terminal moves with the arm swing motion. In the case where the pedestrian is putting the portable terminal in, for example, a pocket of trousers, the traveling direction cannot be determined.

  Note that any of the techniques described in Patent Documents 4 to 7 requires processing for calculating the acceleration in the traveling direction coordinate system from the acceleration data in the sensor coordinate system detected by the sensor. That is, coordinate conversion processing is required for each piece of acceleration data detected by the sensor. This results in an increase in the amount of calculation.

  Therefore, the present invention uses the acceleration sensor and the geomagnetic sensor mounted on the portable terminal even when the pedestrian puts the portable terminal in clothes such as a pants pocket. It is an object of the present invention to provide a portable terminal, a program, and a method for determining a direction as accurately as possible.

According to the present invention, the acceleration sensor that outputs the triaxial acceleration data, the geomagnetic sensor that outputs the triaxial geomagnetic data, and the traveling direction determination means that determines the traveling direction of the pedestrian from the acceleration data and the geomagnetic data. A portable terminal possessed by a pedestrian,
The direction of travel determination means is
The swing period extraction means for extracting the acceleration data of the swing period using each axis acceleration data and the combined acceleration that is the square sum of squares of the acceleration data of all axes,
A specified number of acceleration acquisition means for acquiring a specified number of acceleration data from the acceleration data of the swing phase,
A right / left direction vector calculating means for calculating a right / left direction vector U indicating the right direction or the left direction with respect to the traveling direction of the pedestrian based on the acceleration data of the specified number of swing periods;
It has a traveling direction calculation means for calculating a traveling direction using a right / left direction vector U , a gravity vector G based on acceleration data, and a geomagnetic vector M based on geomagnetic data .

According to another embodiment of the portable terminal of the present invention, the swing phase extraction means is
Detects continuous maximum points from the resultant acceleration,
For the minimum point of the combined acceleration between consecutive local maximum points, it is detected that the combined acceleration continues alternately with “large” and “small”
It is also preferable to detect a point between the maximum points before and after the combined acceleration “small” as the swing leg period.

  According to another embodiment of the mobile terminal of the present invention, the right / left direction vector calculating means is an outer product (U = U) of the acceleration vector B indicating the end of the swing phase and the acceleration vector F indicating the start of the swing phase. It is also preferable to calculate the right / left direction vector U by B × F (or F × B).

According to another embodiment of the mobile terminal of the present invention,
The right / left direction vector calculating means is:
Acceleration surface estimation means for estimating an acceleration surface generated by a walking motion of a pedestrian based on the specified number of swing phase acceleration data;
Vertical acceleration calculation means for calculating vertical acceleration from the specified number of swing phase acceleration data;
It is also preferable to have a longitudinal acceleration calculation means for calculating a normal vector V _LR (right / left direction vector) on the acceleration surface and a longitudinal acceleration from the acceleration surface and the vertical acceleration.

According to another embodiment of the mobile terminal of the present invention,
The longitudinal acceleration calculation means is
Based on the acceleration for each axis output from the acceleration sensor, the gravity vector G is calculated,
From the normal vector V _LR and the gravity vector G on the acceleration surface, a forward or backward front-rear direction vector V _FG is calculated,
Based on the longitudinal vector V _FG and the acceleration indicating the end of the swing leg period, when the longitudinal vector V _FG is forward, the normal vector V _LR is directed right, and when the longitudinal vector V _FG is backward, It is also preferable that the line vector _VLR is leftward.

According to another embodiment of the mobile terminal of the present invention,
The acceleration surface estimation means is
Calculate the variance Vxx, Vyy and Vzz of each axis for the specified number of accelerations,
It is also preferable to estimate the acceleration plane based on the method of least squares for the axis having the minimum variance.

According to another embodiment of the mobile terminal of the present invention,
The right / left direction vector calculation means calculates the right / left by the outer product (U = B × F (or F × B)) of the acceleration vector B indicating the end of the swing phase and the acceleration vector F indicating the start of the swing phase Calculate the leftward vector U,
It is also preferable that the acceleration plane estimation means estimates the acceleration plane based on the least square method for the axis closest to the direction of the right / left direction vector U among the three axes.

According to another embodiment of the mobile terminal of the present invention,
The traveling direction calculation means is
Using the gravity vector G and the geomagnetic vector M, the eastward unit vector e _East is calculated by G × M / | G × M |
Calculate the angle α between the rightward unit vector e _Right and the eastward unit vector e _East in the right / left vector U ,
Based on the angle β formed by the outer product vector (e _East × e _Right ) of the eastward unit vector e _East and the rightward unit vector e _Right and the gravity vector G, the direction angle θ with respect to the eastward unit vector e _East is calculated. It is also preferable.

According to another embodiment of the mobile terminal of the present invention,
About the direction of travel calculation means
The formed angle α is calculated by the following formula:
α = arccos ((e _East · e _Right ) / (| e _East || e _Right |))
The azimuth angle θ is derived by the following equation using the formed angle α, the outer product vector (e _East × e _Right ) and the angle β formed by the gravity vector G: cos β ≧ 0: θ = α
cos β <0: θ = 360−α
It is also preferable.

According to the present invention, a computer mounted on a portable terminal having an acceleration sensor that outputs triaxial acceleration data and a geomagnetic sensor that outputs triaxial geomagnetic data can be used as a pedestrian's progression from acceleration data and geomagnetic data. A program that functions as a traveling direction determination means for determining a direction,
The direction of travel determination means is
The swing period extraction means for extracting the acceleration data of the swing period using each axis acceleration data and the combined acceleration that is the square sum of squares of the acceleration data of all axes,
A specified number of acceleration acquisition means for acquiring a specified number of acceleration data from the acceleration data of the swing phase,
A right / left direction vector calculating means for calculating a right / left direction vector U indicating the right direction or the left direction with respect to the traveling direction of the pedestrian based on the acceleration data of the specified number of swing periods;
Using the right / left direction vector U, the gravity vector G based on the acceleration data, and the geomagnetic vector M based on the geomagnetic data, the computer is caused to function as a traveling direction calculation means for calculating the traveling direction.

According to the present invention, a portable terminal having an acceleration sensor that outputs triaxial acceleration data and a geomagnetic sensor that outputs triaxial geomagnetic data is used to determine a pedestrian's direction of travel from acceleration data and geomagnetic data. A direction determination method,
A first step of extracting the acceleration data of the swing phase using each axis acceleration data and a combined acceleration that is a square sum of squares of the acceleration data of all axes;
A second step of acquiring a specified number of acceleration data from the acceleration data during the swing phase;
A third step of calculating a right / left direction vector U indicating right or left with respect to the direction of travel of the pedestrian based on acceleration data of a specified number of swing periods;
And a fourth step of calculating a traveling direction using a right / left direction vector U , a gravity vector G based on acceleration data, and a geomagnetic vector M based on geomagnetic data .

  According to the portable terminal, the program, and the method of the present invention, even when a pedestrian puts the portable terminal in clothes such as a pants pocket, the acceleration sensor and the geomagnetic sensor mounted on the portable terminal are used. It is possible to determine the traveling direction of the pedestrian as accurately as possible. In particular, when the mobile terminal is worn on the clothes of the lower part of the pedestrian, the traveling direction of the pedestrian can be determined as much as possible.

It is explanatory drawing showing the aspect of the portable terminal put into the pocket of the lower body of a pedestrian. It is a functional block diagram of the portable terminal in this invention. It is a graph showing extraction of a free leg period from each axial acceleration and synthetic acceleration. It is a graph showing transition of the vertical direction acceleration according to elapsed time, and a horizontal direction acceleration. It is a figure showing the acceleration surface and right / left direction vector which actually generate | occur | produce. It is explanatory drawing showing the relationship of the direction with respect to gravity G and geomagnetism M. FIG. It is explanatory drawing showing the relationship between the angle | corner formed and an azimuth angle.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is an explanatory diagram showing an aspect of a mobile terminal placed in a pocket of a pedestrian's lower body.

  According to FIG. 1, the posture of the mobile terminal is not constant when the mobile terminal is inserted into the pocket of the pedestrian's pants. Further, such a state is a case where the user is not operating the mobile terminal and the mobile terminal itself detects the traveling direction by some application. In addition, according to this invention, although the state which the portable terminal was inserted in the pocket of the pants of a pedestrian is assumed, of course, it is not restricted to it. However, according to the present invention, it is preferable that the mobile terminal is located on either the left or right side and located on the lower body portion, particularly when viewed from the center of the pedestrian's body.

  The mobile terminal has a built-in acceleration sensor and geomagnetic sensor fixedly. Therefore, the orientation of the x, y, and z axes (sensor coordinate system) of the sensor is also determined according to the orientation of the mobile terminal. For example, according to FIG. 1, the x-axis is assigned from the top to the bottom of the mobile terminal, the y-axis is from the left to the right of the mobile terminal, and the z-axis is assigned from the back of the mobile terminal to the front. Yes. Of course, the assignment of the coordinate system of the sensor is not limited to this. Further, forward (backward), right (leftward), and vertical coordinate systems are determined for the direction of travel of the pedestrian. Note that the orientation of the portable terminal inserted into the pants pocket cannot be assumed.

  FIG. 2 is a functional configuration diagram of the mobile terminal according to the present invention.

  According to FIG. 2, the mobile terminal 1 is a terminal that can be inserted into a pocket of a user's clothes, for example, a mobile phone or a smartphone. The mobile terminal 1 includes a traveling direction determination unit 10, an acceleration sensor 11, a geomagnetic sensor 12, a positioning unit 13, a map information storage unit 14, a display control unit 15, and a display unit 16. The traveling direction determination unit 10, the map information storage unit 14, and the display control unit 15 are realized by executing a program that causes a computer mounted on the portable terminal to function.

  The acceleration sensor 11 detects acceleration (sensor coordinate system) for each of the x-axis, y-axis, and z-axis. In the case of an existing general mobile phone or smartphone, an acceleration sensor is generally mounted. The detected acceleration data is output to the traveling direction determination unit 10. The acceleration sensor 11 is fixedly incorporated in the mobile terminal 1, and if the orientation of the mobile terminal 1 is determined, the direction of the axis of the acceleration sensor 11 is also determined.

  The geomagnetic sensor 12 detects geomagnetism, which is the magnetic field lines of the earth from south to north. The direction of the orthogonal projection of the detected geomagnetism with respect to the horizontal plane is “north”. In the case of a triaxial geomagnetic sensor, the orientation can be detected by detecting the tilt even if it is not horizontal. The detected geomagnetic data is output to the traveling direction determination unit 10.

  The positioning unit 13 receives a positioning radio wave from a GPS (Global Positioning System) satellite and acquires latitude and longitude data of the current position. The latitude / longitude data is output to the display control unit 15.

  The map information storage unit 14 accumulates map information. Based on the current position instructed from the display control unit 15, the map information is output to the display control unit 15.

  The display control unit 15 receives the traveling direction data from the traveling direction determination unit 10 and the current position information from the positioning unit 13. Further, the display control unit 15 outputs the current position information to the map information storage unit 14 and acquires the map information of the current position. And the display control part 15 produces | generates the image which displayed the present position on the map, and displayed the advancing direction with the arrow. The image is output to the display unit 16. For example, it is used for a navigation system for pedestrians.

  The display unit 16 displays the image from the display control unit 15 so that the walking user can visually recognize the image.

  Next, the traveling direction determination unit 10 which is a characteristic part of the present invention will be described in detail. The traveling direction determination unit 10 includes a swing period extraction unit 101, an acceleration acquisition unit 102 for a specified number, a right / left direction vector estimation unit 103, and a traveling direction calculation unit 104. Hereinafter, these functional components will be described in detail.

<Swing period extraction unit 101>
The free leg period extraction unit 101 extracts acceleration data of the “free leg period” by using the acceleration data of each axis and the combined acceleration that is the square root sum of the acceleration data of all the axes.

  FIG. 3 is a graph showing extraction of the swing leg period from each axis acceleration and combined acceleration.

  As for the acceleration, a small one-step variation and a large one-step variation are alternately observed. According to FIG. 3, the x-axis acceleration, the y-axis acceleration, the z-axis acceleration, and the combined acceleration detected during walking are shown for the portable terminal inserted in the trouser pocket of the pedestrian. The resultant acceleration is the square root of the sum of squares of triaxial acceleration. Note that the posture of the mobile terminal inserted in the pants pocket is unknown.

A human walk is represented by a “walking cycle”. The “walking cycle” refers to the time from the initial contact of the foot on the same side to the next initial contact, and this time is divided into a “standing phase” and a “swing phase”.
“Standing period”: period when the foot is on the ground “Swinging period”: period when the foot leaves the ground and the leg is carried forward by the swing

  For example, suppose that attention is paid to the foot on the pocket side of the pants. Here, it is the “stance phase” that the foot on the pocket side is on the ground, and “the leg phase” is that the leg on the pocket side is carried forward. In the “Standing Period”, the movement of the legs is small, and in the “Standing Period”, the movement of the legs is large. In human walking, the left and right legs are alternately carried forward, so that the “stance phase” and the “free leg phase” are repeated alternately. Therefore, according to the acceleration, a small one-step variation and a large one-step variation are alternately observed.

  The “swing leg period” is between two time points at which the resultant acceleration becomes a maximum point. Moreover, according to FIG. 3, it can be understood that the combined acceleration of the minimum points included in the swing leg period is smaller than the combined acceleration of the minimum points not included in the swing leg period. Therefore, the swing point period is detected by detecting the minimum point of the combined acceleration and detecting the section of the continuous maximum point including the small side where the combined acceleration of the continuous minimum points is small → large.

Specifically, the “swing leg period” is extracted by the following three steps.
(S1) A continuous maximum point is detected from the resultant acceleration.
(S2) For the minimum point of the combined acceleration between successive maximum points, it is detected that the combined acceleration continues alternately “large” and “small”.
(S3) The point between the maximum points before and after the combined acceleration “small” is detected as the “free swing period”.

<Specified number of acceleration acquisition units 102>
The specified number of accelerations acquisition unit 102 acquires the specified number of acceleration data from the acceleration data in the swing phase. Every specified number of acceleration data is output to the right / left direction vector estimation unit 103. The acceleration for the designated number here is preferably about 2 seconds (corresponding to about 4 steps), for example.

<Right / Left Direction Vector Estimation Unit 103>
The right / left direction vector estimation unit 103 estimates the right / left direction vector U based on the specified number of swing phase acceleration data. The “right / left direction vector U” refers to a vector indicating the right side or the left side with respect to the traveling direction of the pedestrian.

Here, the right / left direction vector estimation unit 103 has two processing methods.
[A] Method of calculating right / left direction vector from outer product of acceleration vector of swing leg period [b] Method of calculating right / left direction vector from acceleration plane of acceleration vector of swing leg period

[A] Method for calculating right / left direction vector from outer product of acceleration vectors in swing phase The right / left direction vector estimation unit 103 has only one right / left direction vector calculation unit 103a, and terminates the swing phase period. The right / left direction vector U is calculated from the outer product (U = B × F (or F × B)) of the indicated acceleration vector B and the acceleration vector F indicating the start of the swing phase.

  In general, the normal vector of the acceleration surface in the traveling direction is used to represent the traveling direction of the walking (note that the normal vector of the acceleration surface will be described later in [b] and FIG. 5). A normal vector extending from the origin can be drawn in two directions with respect to the acceleration plane. Since the acceleration plane indicates the traveling direction (before and after the pedestrian), the normal vectors in the two directions correspond to the right and left sides of the pedestrian. Here, of the two normal vectors, the normal vector to the right of the pedestrian is defined as the right vector U. The normal vector on the left side of the pedestrian is defined as a leftward vector U.

Here, the relationship between the acceleration plane and the swing phase is expressed as follows.
Acceleration vector at the top point indicating the front of the acceleration plane
= Acceleration vector F indicating the start of the swing phase
Acceleration vector at the highest point indicating the back of the acceleration plane
= Acceleration vector B indicating the end of the swing leg period
The rightward vector U is calculated by the following equation (1).
Acceleration vector indicating the start of the swing phase: F = (x _F , y _F , z _F )
Acceleration vector indicating the end of the swing leg period: B = (x _B , y _B , z _B )
Right vector indicating the right side of the acceleration surface: U = (x _U , y _U , z _U )
U = B × F
_{= (Y B z F -z B} y F, z B x F -x B z F, x B y F -y B x F) Formula (1)
When the leg is in front, the vector is backwards. Therefore, the right is “foot forward (B) × foot backward (F)”.

When the pedestrian's leftward vector is used as the right / leftward vector, Equation (1) is as follows.
U = F × B
_{= (Y F z B -z F} y B, z F x B -x F z B, x F y B -y F x B)

For example, when the acceleration vectors at the first time and the second time are selected as F and B, respectively, and Expression (1) is applied, the following rightward vector U is obtained.
First time: acceleration of each axis ax = 450, ay = 835, az = 140, combined acceleration a = 959
Second time: acceleration of each axis ax = −369, ay = 980, az = 140, combined acceleration a = 1056
Right vector: U = (20300, 114660, -749115)

[B] Method for calculating right / left direction vector from acceleration surface of acceleration vector in swing phase The right / left direction vector estimation unit 103 includes a vertical direction acceleration calculation unit 103b1, an acceleration surface estimation unit 103b2, and a longitudinal acceleration calculation. Part 103b3.

[Vertical acceleration calculation unit 103b1]
The vertical acceleration calculation unit 103b1 calculates vertical acceleration using the acceleration data. The vertical acceleration and the horizontal acceleration are orthogonal to each other.

  FIG. 4 is a graph showing the transition of vertical acceleration and horizontal acceleration according to elapsed time.

  According to FIG. 4, the pedestrian actually walks with the portable terminal inserted into the pants pocket in the manner as shown in FIG. 1, and the vertical acceleration and horizontal acceleration obtained thereby are plotted. It is.

  According to FIG. 4A, the upper side of the vertical acceleration represents a vertically downward direction, and the lower side represents a vertically upward direction. In addition, the change in vertical acceleration has a periodicity that coincides with walking. Here, the maximum point of vertical downward acceleration represents the time when the body falls, that is, the time when the foot that has left the ground touches the ground. On the other hand, the maximum point of vertical upward acceleration represents the time when the body is raised, that is, the time when the foot is raised. The interval between the minimum points of vertical downward acceleration, that is, the maximum point of vertical upward acceleration, represents one step of the pedestrian.

  According to FIG. 4B, it can be understood that the forward acceleration repeats acceleration / deceleration, and at the same time, repeats forward / backward / upper / lower.

[Acceleration surface estimation unit]
The acceleration surface estimation unit 103b2 estimates an acceleration surface generated by the walking motion of the pedestrian from the specified number of acceleration data.

  FIG. 5 is a diagram showing acceleration planes and right / left-direction vectors that actually occur.

  For the sake of simplicity, the gravity direction is set as the downward direction in the figure, but gravity itself cannot be detected. In addition, in which direction the geomagnetism is actually detected with respect to the sensor coordinate system depends on the attitude of the terminal. However, the relationship between the acceleration plane and the normal vector (right / left vector) does not depend on the attitude of the terminal.

  According to FIG. 5, the three-axis acceleration data (x, y, z) obtained from the acceleration sensor and the three-axis geomagnetic data (x, y, z) obtained from the geomagnetic sensor are expressed in three-dimensional coordinates. Plotted in the system. An acceleration plane can be detected from plots of acceleration data measured at different times. Also, a right / left direction vector with respect to the acceleration plane can be detected.

  According to FIG. 5, the pedestrian is walking in the direction of the direction angle θ with respect to the geomagnetism coming from the south. At this time, the acceleration plane can be detected in parallel with the traveling direction in accordance with the movement of the pedestrian who has inserted the portable terminal into the pants pocket, and the right / left direction vector is specified.

  According to the present invention, a right / left direction vector (normal vector) can be calculated by approximating a plurality of acceleration data to a plane.

As a method for approximately calculating the acceleration surface, for example, there is a least square method. The least square method is a method of determining each coefficient of the prediction function f (x) for the phenomenon so that the sum of the squares of the residuals is minimized. The residual, and the i-th data _{n i} with respect to the predicted function values f _{(n i),} which is the difference between the measured data _{m i,} i.e. _m i -f _{(n i).}

In general, a plane passing through the origin is represented by the following equation (1).
ax + by + cz = 0 Formula (1)
At this time, (a, b, c) is a normal vector with respect to the plane.

Here, in order to simplify the calculation, Equation (1) is transformed into Equation (2).
z = αx + βy Equation (2)
Since (a, b, c) is a normal vector, there is no problem when c = -1.

When n point groups x _i , y _i , and z _i (i = 1 to n) are given, α and β that minimize the following expression (3) may be calculated.
S = Σ _{i = 1} ⁿ (z _i −αx _i −βy _i ) Equation ² (3)

Here, it is assumed that it is defined as follows.
A = Σ _{i = 1} ⁿ (x _i ² )
B = Σ _{i = 1} ⁿ (y _i ² )
C = Σ _{i = 1} ⁿ (z _i ² )
D = Σ _{i = 1} ⁿ (x _i × y _i )
E = Σ _{i = 1} ⁿ (x _i × z _i )
F = Σ _{i = 1} ⁿ (y _i × z _i )

At this time, the expression (3) becomes the following expression (4).
S = Aα ² + Bβ ² + C + 2αβD-2αE-2βF Formula (4)

When Expression (4) is regarded as a function of α, it becomes a concave quadratic function, and the minimum point value is minimized. The same applies to β. That is, S is partially differentiated with respect to α and β, and a point that becomes 0 is obtained.

When the equations (5) and (6) are solved, they can be calculated as the following equations (7) and (8).
α = (BE-DF) / (AB-D ² ) Formula (7)
β = (AF−DE) / (AB−D ² ) Formula (8)
As described above, the normal vector is (α, β, −1).

  Further, as another embodiment for estimating the acceleration plane, according to the above-described equation (2), the acceleration plane is calculated so that the residual in the z-axis direction is minimized. However, it is preferable to select an axis with a small variance as the axis that minimizes the residual. For example, there are the following two methods for selecting the axis that minimizes the residual.

(1) For the specified number of accelerations, variances Vxx, Vyy and Vzz of each axis are calculated, and the acceleration plane is estimated based on the least square method for the axis having the smallest variance.
_{Vxx = 1 / n · Σ i} = 1 n (x i -x¯) 2
_{Vyy = 1 / n · Σ i} = 1 n (y i -y¯) 2
Vzz = 1 / n · Σ _{i = 1} ⁿ (z _i −z￣) ²
n: Number of data x￣: Average value of x

(2) The acceleration plane is estimated based on the least square method for the axis closest to the direction of the right / left direction vector U among the three axes. Here, the right vector is calculated by the same processing as the right / left vector calculation unit 103a. For example, in the case of the graph of FIG. 4 described above, the following numerical values are obtained.
Acceleration vector indicating the start of the swing phase: F = (-1.50, 0.15, -0.14)
Acceleration vector indicating the end of the swing phase: B = (-1.06, 0.21, 0.69)
Right vector indicating the right side of the acceleration plane: U = (-0.13, -1.18, 0.16)
In this case, the y axis is selected as the axis closest to the direction of the right vector.

  Note that the acceleration surface estimation unit 103b2 also preferably functions as a low-pass filter that blocks high frequency components with respect to acceleration data. As a result, acceleration data as an abnormal value can be removed, and an accurate traveling direction can be calculated.

[Front-back direction acceleration calculation unit]
The longitudinal acceleration calculation unit 103b3 calculates a right / left direction vector (normal vector) and a longitudinal acceleration on the acceleration surface from the acceleration surface and the vertical acceleration. Specifically, the following steps S1 to S3 are included.

(S1) The longitudinal acceleration calculation unit 103b3 first calculates the gravity vector G based on the acceleration for each axis output from the acceleration sensor.

  FIG. 6 is an explanatory diagram showing the orientation relationship with respect to gravity G and geomagnetism M. FIG.

The gravity vector G and the geomagnetic vector M are expressed as follows for each of the x-axis, the y-axis, and the z-axis. Although the posture of the mobile terminal while walking is not necessarily constant, the geomagnetic data observed in a short time is almost constant.
Gravity vector G: G = (g _x , g _y , g _z )
Geomagnetic vector _{M: M = (m x,} m y, m z)

  First, the gravity vector G is calculated. It is difficult to accurately determine the gravitational direction of the mobile terminal during walking even when using acceleration data. Therefore, the direction of the vector of the sum of the accelerations of each axis is regarded as the direction of gravity using a large number of x-axis, y-axis and z-axis acceleration data detected in a predetermined time range.

The acceleration of each i-th axis is expressed as follows.
x-axis acceleration: ACC _x [i]
y-axis acceleration: ACC _y [i]
z-axis acceleration: ACC _z [i]
The sum of n pieces of acceleration data is expressed as follows.
ACCS _x = Σ _{i = 1} ^N ACC _x [i]
ACCS _y = Σ _{i = 1} ^N ACC _y [i]
ACCS _z = Σ _{i = 1} ^N ACC _z [i]
The gravity vector G is expressed as follows.
G = (g _x , g _y , g _z ) = (ACCS _x , ACCS _y , ACCS _z )

(S2) Next, the longitudinal acceleration calculation unit 103b3 calculates a forward or backward longitudinal vector V _FG from the normal vector V _LR and the gravity vector G of the acceleration surface. Here, the normal vector V _LR and the gravity vector G of the acceleration surface and the longitudinal vector V _FG are expressed by the following relational expressions.
V _FG = V _LR × G ×: Cross product

A unit vector in the front-rear direction vector V _FG is expressed as follows.
e _FB = (e _FBx , e _FBy , e _FBz )
At this time, the longitudinal acceleration A _FG is expressed as follows.
A _FG = e _FGx × A _x + e _FGy × A _z + e _FGz × A _z
A _FG is a forward or backward acceleration.

(S3) Next, longitudinal acceleration calculating unit 103b3 based on the acceleration indicated by the end of the longitudinal direction vector V _FG and swing phase, to determine the orientation of the right vector (normal vector).
Longitudinal vector V _FG : Forward-> Normal vector V _LR : Right (+)
Front-rear direction vector V _FG : Backward-> Normal vector V _LR : Leftward (-)

Whether forward or backward, the direction of acceleration indicating the end of the swing phase can be determined as “rear”. The pedestrian's foot during the swing phase is carried from back to front. Therefore, the portable terminal inserted in the pants pocket moves while drawing a circular arc in a pendulum shape. Therefore, when the mobile terminal is in front, the acceleration is detected backward in the front-rear direction. Incidentally, among the three axes, the axis that is closest to the direction of the right / left vector U, when estimating the acceleration surface based on the least square method, based on the law line vector V _LR, right / left vector ( The direction of the normal vector).

<Advancing direction calculation unit 104>
The traveling direction calculation unit 104 calculates the traveling direction using the right / left direction vector U, the gravity vector G, and the geomagnetic vector M. The calculated traveling direction is output to the display control unit 15. Specifically, the following steps S1 to S3 are included.

(S1) First, the traveling direction calculation unit 104 calculates the eastward unit vector e _East by G × M / | G × M | using the gravity vector G and the geomagnetic vector M.

  Reference is again made to FIG.

Next, assuming that the x-axis, y-axis, and z-axis are right-handed, for example, the unit vector in each direction is expressed as follows.
North-facing unit vector e _North : e _North = (e _Nx , e _Ny , e _Nz )
Eastward unit vector e _East : e _East = (e _Ex , e _Ey , e _Ez )
Downward unit vector e _Down : e _Down = (e _Dx , e _Dy , e _Dz )
Further, the acceleration data is expressed as follows.
A = (A _x , A _y , A _z )

According to FIG. 6, when the gravity vector G and the geomagnetic vector M are used, each unit vector is expressed as follows.
North-facing unit vector e _North : e _North = G × M × G / | G × M × G |
East unit vector e _East : e _East = G × M / | G × M |
Downward unit vector e _Down : e _Down = G / | G |
×: Cross product (vector product, outer product)

  “G × M” (outer product) creates a vector in a direction perpendicular to the plane formed by the gravity vector G and the geomagnetic vector M. When viewed in the direction of the gravity vector G, G × M extends in a clockwise direction as viewed from the geomagnetic vector M. Since the geomagnetic vector M is directed from the south to the north, the direction of 90 degrees clockwise from the axis, that is, G × M is directed to the “east” direction.

  Also, “(G × M) × G” creates a vector in a direction perpendicular to the plane created by the east vector G × M and the gravity vector G. When looking toward the east vector G × M, (G × M) × G extends in the clockwise direction as viewed from the gravity vector. Since the gravity vector is directed from the top to the bottom, the direction of 90 degrees clockwise from the axis, that is, (G × M) × G is directed to the “north” direction.

The acceleration vector in each direction is calculated by taking the inner product of the acceleration vector A with respect to the unit vector e in each direction.
Northward acceleration: A _N = e _Nx × A _x + e _Ny × A _y + e _Nz × A _z
East acceleration: A _E = e _Ex × A _x + e _Ey × A _y + e _Ez × A _z
Vertical downward acceleration: A _D = e _Dx × A _x + e _Dy × A _y + e _Dz × A _z

(S2) Next, the traveling direction calculation unit 104 calculates an angle α formed by the rightward unit vector e _Right and the eastward unit vector e _East in the normal vector. The angle formed by the two vectors can be calculated from the inner product and the norm (vector magnitude) product. From the eastward unit vector e _East and the rightward unit vector e _Right , the angle α formed by both vectors is expressed by the following equation.
α = arccos ((e _East · e _Right ) / (| e _East || e _Right |)) Equation (9)

(S3) Then, the traveling direction calculation unit 104 obtains the azimuth angle θ.

  Here, the formed angle α is an inferior angle (0 to 180 degrees). On the other hand, the azimuth angle θ to be obtained is 0 to 360 degrees (clockwise angle with the north azimuth being 0 degrees) and includes a dominant angle. Therefore, the formed angle is converted into an azimuth angle.

  FIG. 7 is an explanatory diagram showing the relationship between the formed angle and the azimuth angle.

According to FIG. 7, according to the outer product vector e _East × e _Right of the east unit vector e _East and the right unit vector e _Right , the same as the gravity vector when the direction angle θ with respect to the east unit vector e _East is less than 180. Indicates direction. At this time, θ = α. On the other hand, when θ should be 180 degrees or more, e _East × e _Right indicates a direction opposite to the gravity vector. At this time, θ = 360−α. Therefore, an angle formed by e _East × e _Right and the gravity vector G is obtained (cosine instead of angle), and α is converted to θ based on the result.

The direction angle θ with respect to the eastward unit vector e _East is derived by the following equation using β as the angle formed by e _East × e _Right and the gravity vector G.
cos β ≧ 0: θ = α
cos β <0: θ = 360−α

  As described above in detail, according to the mobile terminal, the program, and the method of the present invention, even if the pedestrian puts the mobile terminal in clothes such as a pants pocket, The traveling direction of the pedestrian can be determined as accurately as possible using the mounted acceleration sensor and geomagnetic sensor. In particular, when the mobile terminal is worn on the clothes of the lower part of the pedestrian, the traveling direction of the pedestrian can be determined as much as possible.

  Various changes, modifications, and omissions of the above-described various embodiments of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 Mobile terminal 10 Travel direction determination part 101 Swing leg period extraction part 102 Specified number of acceleration acquisition part 103 Right / left direction vector estimation part 103a Right / left direction vector calculation part 103b1 Vertical direction acceleration calculation part 103b2 Acceleration surface estimation part 103b3 Longitudinal direction acceleration Calculation unit 104 Traveling direction calculation unit 11 Acceleration sensor 12 Geomagnetic sensor 13 Positioning unit 14 Map information storage unit 15 Display control unit 16 Display unit

Claims (11)

An acceleration sensor that outputs triaxial acceleration data; a geomagnetic sensor that outputs triaxial geomagnetic data; and a traveling direction determination means that determines a traveling direction of a pedestrian from the acceleration data and the geomagnetic data. A portable terminal owned by a person,
The traveling direction determination means includes
The swing period extraction means for extracting the acceleration data of the swing period using each axis acceleration data and the combined acceleration that is the square sum of squares of the acceleration data of all axes,
A specified number of acceleration acquisition means for acquiring a specified number of acceleration data from the acceleration data of the swing phase,
A right / left direction vector calculating means for calculating a right / left direction vector U indicating the right direction or the left direction with respect to the traveling direction of the pedestrian based on the acceleration data of the specified number of swing periods;
A portable terminal comprising: a traveling direction calculation means for calculating a traveling direction using the right / left direction vector U , a gravity vector G based on the acceleration data, and a geomagnetic vector M based on the geomagnetic data. . The swing leg extraction means includes
From the synthesized acceleration, a continuous maximum point is detected,
For the minimum point of the composite acceleration between the continuous maximum points and the maximum point, it is detected that the composite acceleration continues alternately with “large” and “small”,
The mobile terminal according to claim 1, wherein a portion between local maximum points before and after the combined acceleration “small” is detected as a swing leg period.   The right / left direction vector calculating means is based on an outer product (U = B × F (or F × B)) of an acceleration vector B indicating the end of the swing phase and an acceleration vector F indicating the start of the swing phase. The mobile terminal according to claim 2, wherein a right / left vector U is calculated. The right / left direction vector calculating means includes:
Acceleration surface estimation means for estimating an acceleration surface generated by a walking motion of a pedestrian based on the specified number of swing phase acceleration data;
Vertical acceleration calculation means for calculating vertical acceleration from the specified number of swing phase acceleration data;
A longitudinal acceleration calculation means for calculating a normal vector V _LR (right / left direction vector U) and a longitudinal acceleration on the acceleration surface from the acceleration surface and the vertical acceleration, respectively. 2. The mobile terminal according to 2. The longitudinal acceleration calculation means includes:
Based on the acceleration for each axis output from the acceleration sensor, the gravity vector G is calculated,
From the normal vector V _LR and the gravity vector G on the acceleration surface, a forward or backward front-rear direction vector V _FG is calculated,
Based on the front-rear direction vector V _FG and the acceleration indicating the end of the swing leg period, when the front-rear direction vector V _FG is forward, the normal vector V _LR is directed right, and the front-rear direction vector V _FG is rearward 5. The mobile terminal according to claim 4, wherein the normal vector V _LR is leftward. The acceleration surface estimation means includes
Calculate the variance Vxx, Vyy and Vzz of each axis for the specified number of accelerations,
The mobile terminal according to claim 4 or 5, wherein an acceleration plane is estimated based on a least square method for an axis having a minimum variance. The right / left direction vector calculating means is based on an outer product (U = B × F (or F × B)) of an acceleration vector B indicating the end of the swing phase and an acceleration vector F indicating the start of the swing phase. , Calculate the right / left vector U,
6. The acceleration surface estimating means estimates an acceleration surface based on a least square method for an axis closest to the direction of the right / left direction vector U among three axes. Mobile devices. The traveling direction calculation means includes
Using the gravity vector G and the geomagnetic vector M, an eastward unit vector e _East is calculated by G × M / | G × M |
Calculating an angle α between a rightward unit vector e _Right in the right / leftward vector U and the eastward unit vector e _East ;
Based on the angle β formed by the outer product vector (e _East × e _Right ) of the eastward unit vector e _East and the rightward unit vector e _Right and the gravity vector G, the direction angle θ with respect to the eastward unit vector e _East The mobile terminal according to claim 1, wherein the mobile terminal is calculated. About the traveling direction calculation means,
The formed angle α is calculated by the following equation:
α = arccos ((e _East · e _Right ) / (| e _East || e _Right |))
The azimuth angle θ is derived by the following equation using the formed angle α, the outer product vector (e _East × e _Right ) and the angle β formed by the gravity vector G: cos β ≧ 0: θ = α
cos β <0: θ = 360−α
The mobile terminal according to claim 8. A computer mounted on a portable terminal having an acceleration sensor that outputs triaxial acceleration data and a geomagnetic sensor that outputs triaxial geomagnetic data is used to determine the direction of travel of a pedestrian from the acceleration data and the geomagnetic data. A program that functions as a direction of travel determination means,
The traveling direction determination means includes
The swing period extraction means for extracting the acceleration data of the swing period using each axis acceleration data and the combined acceleration that is the square sum of squares of the acceleration data of all axes,
A specified number of acceleration acquisition means for acquiring a specified number of acceleration data from the acceleration data of the swing phase,
A right / left direction vector calculating means for calculating a right / left direction vector U indicating the right direction or the left direction with respect to the traveling direction of the pedestrian based on the acceleration data of the specified number of swing periods;
A computer is caused to function as a traveling direction calculation means for calculating a traveling direction using the right / left direction vector U , a gravity vector G based on the acceleration data, and a geomagnetic vector M based on the geomagnetic data. A program for mobile devices. A traveling direction determining method for determining a traveling direction of a pedestrian from the acceleration data and the geomagnetic data by a portable terminal having an acceleration sensor that outputs triaxial acceleration data and a geomagnetic sensor that outputs triaxial geomagnetic data. There,
A first step of extracting the acceleration data of the swing phase using each axis acceleration data and a combined acceleration that is a square sum of squares of the acceleration data of all axes;
A second step of acquiring a specified number of acceleration data from the acceleration data of the swing phase;
A third step of calculating a right / left direction vector U indicating right or left with respect to the direction of travel of the pedestrian based on acceleration data of a specified number of swing periods;
4. A portable terminal comprising: a fourth step of calculating a traveling direction using the right / left direction vector U , a gravity vector G based on the acceleration data, and a geomagnetic vector M based on the geomagnetic data. How to determine the direction of travel.

Images