JP4793223B2 - Walking navigation method, system and program - Google Patents

Description

  The present invention relates to a walking navigation method, system and program, and more particularly to a walking navigation method based on autonomous navigation.

  Patent Documents 1, 2, and 3 disclose techniques for specifying the current location by autonomous navigation in a portable information terminal. In a navigation system mounted on a vehicle, the relationship between the posture of a vehicle that is a moving body and the posture of a magnetic sensor is fixed, whereas in a portable information terminal, a magnetic sensor is used for the posture of a pedestrian that is a moving body. The posture is not stable. For this reason, in a portable information terminal, it is not easy to derive the walking direction and derive the current location by autonomous navigation using the walking direction.

  According to the method disclosed in Patent Document 1, the horizontal component of acceleration at the moment when the differential value of the horizontal component of acceleration of the three-dimensional acceleration sensor becomes maximum for each step based on the acceleration data output from the three-dimensional acceleration sensor. The direction of the horizontal component of the acceleration at the moment when the horizontal component of the acceleration is maximized for each step is derived as the walking direction of the section that takes one step.

JP 2003-302419 A JP-A-2005-283386 Japanese Patent Laid-Open No. 2-216011

  However, the method disclosed in Patent Document 1 has a problem that errors are likely to accumulate in order to derive the walking direction for each step. For example, when the time interval for sampling acceleration is wide, the difference between the true walking direction and the derived walking direction becomes large. However, when the time interval for sampling acceleration is narrow, the amount of calculation increases, so the time interval for sampling cannot be reduced to a dark cloud.

  Moreover, in the method disclosed in Patent Document 1, the inclination of the acceleration sensor is derived in a state where the acceleration sensor is stationary, and the walking direction is derived for each step by using the inclination of the acceleration sensor thus derived. . However, since there is no guarantee that the posture of the acceleration sensor in the stationary state matches the posture of the acceleration sensor in the walking state, the estimation error in the walking direction may increase beyond the allowable range in the method disclosed in Patent Document 1.

  The present invention was created in view of such a problem, and an object thereof is to improve the accuracy of walking navigation by autonomous navigation.

  (1) A walking navigation method for achieving the above object is a discrete navigation method in which an acceleration data is output from a three-dimensional acceleration sensor that outputs acceleration data as data indicating a combined vector of gravity and inertial force acting on a unit mass. Obtaining the horizontal component and vertical axis component of the acceleration of the three-dimensional acceleration sensor based on the acceleration data, and obtaining the three-dimensional horizontal vibration axis that is the major axis of the horizontal component distribution of the acceleration An inclination based on the attitude of the acceleration sensor is derived by statistical calculation, a horizontal vibration axis component of the acceleration is derived based on the inclination of the horizontal vibration axis, and a vibration phase of the horizontal vibration axis component of the acceleration and the acceleration One of two directions parallel to the horizontal vibration axis based on the deviation of the vertical axis component from the vibration phase of the vertical axis component. Determining the orientation in degrees sensor as walking direction as a reference, it comprises.

  The acceleration data output from the three-dimensional acceleration sensor is three-dimensional vector data. By defining an arbitrary three-axis orthogonal coordinate system that rotates together with the three-dimensional acceleration sensor with respect to the acceleration data, the direction of acceleration based on the attitude of the three-dimensional acceleration sensor is determined. The arbitrary section is an arbitrary section that may be partitioned by time, separated by distance, or separated by the number of steps. The horizontal component of the acceleration of the three-dimensional acceleration sensor is two-dimensional vector data that is processed by regarding it as the horizontal component of the acceleration of the three-dimensional acceleration sensor. The vertical axis component of the acceleration of the three-dimensional acceleration sensor is one-dimensional vector data that is processed by regarding it as the vertical axis component of the acceleration of the three-dimensional acceleration sensor. Since the vertical axis component of acceleration is one-dimensional vector data that is processed as “perceived” as the vertical axis component of acceleration, it may be data equivalent to the acceleration of a three-dimensional acceleration sensor, It may be data corresponding to a combined vector of gravitational acceleration. The method for deriving the horizontal component and the vertical axis component of the acceleration of the three-dimensional acceleration sensor based on the acceleration data is not limited, and for example, the method described in Patent Document 1 may be used. In the present specification, the direction and direction are properly used as follows. The direction refers to a direction defined in an arbitrary coordinate system determined in advance, and the direction refers to a direction based on the four directions of east, west, south, and north. Therefore, the “walking direction based on the posture of the three-dimensional acceleration sensor” is a direction defined in a coordinate system in which the coordinate axis rotates together with the three-dimensional acceleration sensor.

  In this walking navigation method, the walking direction is derived by statistical calculation based on a plurality of acceleration data acquired from the three-dimensional acceleration sensor in an arbitrary section. For this reason, the walking direction can be derived with higher accuracy than the method of deriving the walking direction for each step by appropriately setting the length of the arbitrary section that divides the population. That is, in the method of deriving the walking direction for each step as disclosed in Patent Document 1, since the walking direction is determined for each step, errors determined each time the walking direction is determined accumulates. On the other hand, in the walking navigation method according to the present invention, acceleration data divided by an arbitrary section that advances, for example, 5 steps and 10 steps is statistically processed, so that an error in the derived walking direction is reduced.

  In this walking navigation method, the horizontal vibration axis derived by statistical calculation does not inherently have a unique direction, so the data derived as the inclination of the horizontal vibration axis based on the posture of the three-dimensional acceleration sensor is parallel to the walking direction. Even if it is a value that can uniquely identify a specific direction, it may indicate a walking direction, or may indicate a direction opposite to the walking direction. When the data derived as the inclination of the horizontal vibration axis indicates the walking direction, the horizontal vibration axis component of the acceleration of the acceleration sensor coincides with the walking direction axis component of the acceleration of the acceleration sensor. The walking direction axis here is a coordinate axis having a direction in which the walking direction is positive. In the latter case, the horizontal vibration axis component of the acceleration of the acceleration sensor has the same magnitude as that of the walking direction axis component of the acceleration of the acceleration sensor and has an opposite relationship. In the present invention, focusing on the phenomenon that the vibration phase of the walking direction axis component of the acceleration of the acceleration sensor carried by the pedestrian deviates from the vibration phase of the vertical axis component of the acceleration, one of the two directions parallel to the horizontal vibration axis is selected. One direction is determined as the walking direction. Of course, neither the vibration of the walking direction axis component of the acceleration of the acceleration sensor carried by the pedestrian nor the vibration of the vertical axis component is a perfect periodic function, but it is one step in both the walking direction axis component vibration and the vertical axis component vibration A similar vibration pattern appears every time, and the vibration phase of the walking direction axis component of the acceleration of the three-dimensional acceleration sensor slightly advances with respect to the vibration phase of the vertical axis component. Therefore, for example, when the vibration phase of the vertical axis component advances relative to the vibration phase of the horizontal vibration axis component, it is determined that the data derived as the inclination of the horizontal vibration axis indicates a direction opposite to the walking direction. The walking direction can be determined by an angle obtained by adding 180 ° to the angle representing the inclination of the axis. In the opposite case, it can be determined that the data derived as the inclination of the horizontal vibration axis indicates the walking direction, and the walking direction can be determined by the angle representing the inclination of the horizontal vibration axis.

  (2) In the walking navigation method for achieving the above object, the direction of stationary data regarded as acceleration data output from the stationary three-dimensional acceleration sensor from the coordinate system unique to the three-dimensional acceleration sensor is Z-axis A mapping to an XYZ Cartesian coordinate system that is parallel may be derived. In this case, the horizontal component of the acceleration is an X, Y component of the acceleration data in the XYZ orthogonal coordinate system, and the vertical axis component of the acceleration is a Z component of the acceleration data in the XYZ orthogonal coordinate system. It is.

  The force acting on the mass in the uniform linear motion on the earth's surface is gravity. The three-dimensional acceleration sensor that outputs the data indicating the combined vector of gravity and inertial force that acts on the unit mass as acceleration data is a three-dimensional equivalent to the acceleration in the direction opposite to the gravitational acceleration. Output as vector data. That is, the static data is data that is processed assuming that it corresponds to an acceleration whose direction is opposite to that of the gravitational acceleration. There is no restriction | limiting in particular in the method of deriving static data, For example, you may use the method described in patent document 1. FIG.

  In this walking navigation method, the coordinate axis of acceleration data is rotated so that the direction of still data to be processed is parallel to the Z axis, assuming that the acceleration corresponds to the acceleration opposite to the gravitational acceleration. That is, when still data is expressed in the XYZ orthogonal coordinate system, both the X component and the Y component of acceleration data are zero. By handling the acceleration data by rotating the coordinate axes in this way, the acceleration data can be handled easily.

(3) The acceleration data output from the three-dimensional acceleration sensor during the constant-velocity linear motion on flat ground is three-dimensional vector data equivalent to gravitational acceleration. A straight line walk on flat ground can be approximated by a uniform linear motion on flat ground. Curve walking on flat ground can be regarded as straight walking on flat ground in a short time interval. Therefore, if a section having an appropriate length is set, walking on a flat ground can be approximated to a uniform linear motion on a flat ground.
Therefore, in the walking navigation method for achieving the above object, an average of a plurality of the acceleration data may be derived as the stationary data.

  In this method, stationary data can be derived in a walking state. Here, the difference between the direction of the static data derived based on the acceleration data acquired during walking and the opposite direction of the gravitational acceleration, and the direction of the static data derived based on the acceleration data acquired while stationary and the gravitational acceleration The deviation from the opposite direction will be described. The deviation between the opposite direction of the stationary data and the direction of gravitational acceleration corresponds to the deviation between the true posture of the three-dimensional acceleration sensor and the estimated posture. If the static data is derived based on the acceleration data acquired while stationary, the difference between the static data and the opposite direction of the gravitational acceleration is compared to the case where the static data is derived based on the acceleration data acquired while walking. Become smaller. However, there is no guarantee that the posture of the three-dimensional acceleration sensor in a stationary state matches the posture of the three-dimensional acceleration sensor during walking. If the posture of the acceleration sensor during walking is estimated based on the stationary data derived in the stationary state, the posture estimation error increases depending on the posture difference between the stationary state and the walking state of the acceleration sensor. Estimation error may be unacceptably large. As a result, the walking direction derived according to the posture of the acceleration sensor estimated from the stationary state tends to include an unacceptable error with respect to the actual walking direction. On the other hand, when the average of acceleration data acquired during walking is derived as stationary data, the orientation of the 3D acceleration sensor corresponding to the stationary data derived as such and the true orientation of the 3D acceleration sensor This difference is only an error resulting from approximating the motion of the acceleration sensor during walking to a constant velocity linear motion, and therefore it is difficult to include an unacceptable error. Therefore, according to the present invention, the walking direction derived from the inclination of the horizontal vibration axis is unlikely to include an unacceptable error with respect to the actual walking direction.

(4) In the walking navigation method for achieving the above object, the difference between the vibration phase of the horizontal vibration axis component of the acceleration and the vibration phase of the vertical axis component of the acceleration is expressed as the horizontal vibration axis component of the acceleration. The ratio of the vibration phase delay of the vertical axis component of the acceleration to the vibration phase of the acceleration and the period of the horizontal vibration axis component of the acceleration or the vibration period of the vertical axis component of the acceleration may be used.
In addition, since the vibration period of the horizontal vibration axis component of acceleration and the vibration period of the vertical axis component of acceleration coincide with each other in the walking state, no matter what the vibration period is compared with the delay of the vibration phase, it is substantial. There is no difference.

(5) In the walking navigation method for achieving the above object, magnetic data is acquired from a magnetic sensor that changes posture with the three-dimensional acceleration sensor, and the walking direction in the arbitrary section is based on the magnetic data and the walking direction. May be derived.
In this case, it is possible to derive the walking direction based on the four directions of east, west, south, and north.

(6) As described above, in the walking state, the vibration of the vertical axis component of acceleration is a periodic motion in which a similar vibration pattern appears at every step. Therefore, the number of steps can be derived based on the vertical axis component of acceleration.
Therefore, in the walking navigation method for achieving the above object, the number of steps in the arbitrary section is derived based on the vertical axis component of the acceleration, and the walking distance of the arbitrary section is determined based on the reference stride and the number of steps in the arbitrary section. May be derived.

In this case, it is possible to derive the relative position of the end point of the arbitrary section based on the starting point of the arbitrary section based on the walking direction and the walking distance for each arbitrary section. In other words, by deriving the relative position of the end point with respect to the start point for each arbitrary section and accumulating the results, it is possible to derive the current location based on the point where the position is specified by GPS position data etc. Become.
The reference stride is a value used as an average stride in an arbitrary section in order to estimate the walking distance from the number of steps, and may be derived by calculation or registered by the user.

  (7) For example, in the walking navigation method for achieving the above object, the section distance from the step count calculation location to the step count calculation location is derived based on the GPS position data of the step count calculation location and the GPS position data of the step count settlement location And deriving a reference number of steps from the step count value to the step count settlement value based on the vertical axis component of the acceleration, and deriving the reference step length based on the section distance and the reference step number.

The step count location is the starting point of a section that is provisionally set in order to derive the stride based on GPS position data. The step count settlement point is an end point of a section that is provisionally set in order to derive a stride based on GPS position data.
In this case, the labor for registering basic data such as height for deriving the reference stride and the reference stride can be saved, so that the usability is improved.

(8) In the walking navigation method for achieving the above object, the current position may be derived based on the walking direction, the walking distance, and GPS position data.
In this case, the current position of the pedestrian can be derived by highly accurate autonomous navigation even in an arbitrary section where the current position cannot be specified by satellite navigation.
(9) In the walking navigation method for achieving the above object, the horizontal component of the acceleration in the arbitrary section is (X _p , Y _p ), where (p = 0, 1, 2,... S). At this time, θ (−90 ° <θ ≦ 90 °) satisfying the following expression (1) may be derived as the inclination of the horizontal vibration axis.

Ｘ軸の回転後の座標軸ｔが水平振動軸となるθを導出するためには、次式（３）で表すｆ（θ）が極大となるθを求めればよい。

A mapping for rotating the XY coordinate axis by θ in the same coordinate plane and converting it to the tu coordinate axis is expressed by the following equation (2).

In order to derive θ where the coordinate axis t after rotation of the X-axis becomes the horizontal vibration axis, θ that maximizes f (θ) expressed by the following equation (3) may be obtained.

  Equation (1) is satisfied by two θ values at which f (θ) takes an extreme value. One of the two θs satisfying the expression (1) has f (θ) as a maximum, and the other has f (θ) as a minimum. Therefore, of the two θs satisfying the expression (1), when f (θ) is larger when f (θ) is substituted, the value of f (θ) becomes a maximum. The method of using θ derived in this way as the inclination of the horizontal vibration axis in the XY orthogonal coordinate system is one method of deriving the inclination of the horizontal vibration axis by statistical calculation.

(10) A walking navigation system for achieving the above-described object is a discrete navigation system in an arbitrary section from a three-dimensional acceleration sensor that outputs data indicating a combined vector of gravity and inertial force acting on a unit mass as acceleration data. Means for obtaining in time; means for deriving a horizontal component and a vertical axis component of acceleration of the three-dimensional acceleration sensor based on the acceleration data;
Means for deriving, by statistical calculation, an inclination of the horizontal vibration axis, which is the major axis of the horizontal component distribution of the acceleration, based on the attitude of the three-dimensional acceleration sensor; and a vibration phase of the horizontal vibration axis component of the acceleration; Based on the deviation of the acceleration from the vibration phase of the vertical axis component, one of two directions parallel to the horizontal vibration axis is set as a walking direction based on the posture of the three-dimensional acceleration sensor in the arbitrary section. And means for determining.

  (11) A walking navigation program for achieving the above object is a discrete navigation program in an arbitrary section from a three-dimensional acceleration sensor that outputs data indicating a combined vector of gravity and inertial force acting on a unit mass as acceleration data. Means for obtaining time; means for deriving a horizontal component and a vertical axis component of the acceleration of the three-dimensional acceleration sensor based on the acceleration data; and a horizontal vibration axis that is a major axis of the distribution of the horizontal component of the acceleration Based on a deviation between a vibration phase of the horizontal vibration axis component of the acceleration and a vibration phase of the vertical axis component of the acceleration. A walking method based on the posture of the three-dimensional acceleration sensor in the arbitrary section in any one of two directions parallel to the horizontal vibration axis Determining means and causes a computer to function as a.

Hereinafter, embodiments of the present invention will be described in the following order with reference to the accompanying drawings.
********table of contents*************
(1) Basic principle (2) Hardware configuration of walking navigation system (3) Data flow of walking navigation system (4) Details of walking direction derivation method-Estimation of acceleration sensor tilt-Horizontal component of acceleration and vertical axis Derivation of components-Derivation of the inclination of the horizontal vibration axis-Derivation of walking direction-Derivation of walking direction (5) Effect (6) Other Embodiments *************** ＊＊＊＊＊

(1) Basic Principle First, the basic principle for facilitating understanding of the embodiment of the present invention will be described with reference to FIGS.
The x, y, and z axes in FIG. 1 indicate coordinate axes unique to the acceleration sensor and the magnetic sensor. In FIG. 1, the acceleration sensor and the magnetic sensor are relatively fixed, and the direction of each coordinate axis unique to the acceleration sensor coincides with the direction of each coordinate axis unique to the magnetic sensor (the corresponding coordinate axis It is assumed that the directions match and the variables taken on the corresponding coordinate axes match.)

  When the discrete time output of an acceleration sensor carried by a pedestrian who is walking in a straight line is plotted on the xyz coordinate system for a certain time interval, the output shows a three-dimensional finite distribution. The acceleration sensor carried by a pedestrian moves three-dimensionally with the movement of the body part where the acceleration sensor is located, but the movement that occurs during linear walking on a horizontal plane is two-dimensional. That is, acceleration data that is an output of the acceleration sensor is distributed in the vicinity of a plane including the vertical axis and the walking direction axis perpendicular thereto. When this plane is called a walking vibration plane, the walking vibration plane is a plane perpendicular to the horizontal plane. When the acceleration data is decomposed into a vertical axis component that is a one-dimensional vector and a horizontal component that is a two-dimensional vector, and the horizontal component is further decomposed into a component parallel to the walking vibration surface and a component perpendicular to the walking vibration surface, the acceleration data The horizontal component parallel to the walking vibration plane is parallel to the walking direction axis.

  Considering the distribution of data obtained by projecting acceleration data perpendicular to the horizontal plane (the horizontal plane is indicated by a circle in FIG. 1), that is, the distribution of the horizontal component of the acceleration data, the walking direction axis included in the horizontal plane The variance of the distribution is maximized in the direction, and the variance is minimized in the direction perpendicular to the walking direction axis. That is, the direction in which the dispersion of the XY component of the acceleration data corresponding to the data obtained by projecting the acceleration data perpendicular to the horizontal plane is parallel to the walking direction axis.

  Based on the above, if acceleration data is treated as spatial information and the inclination of the vertical axis with respect to the coordinate axis unique to the acceleration sensor can be estimated as the inclination of the acceleration sensor, statistical calculation using the data obtained by projecting the acceleration data on the estimated horizontal plane as the population It can be seen that the walking direction axis can be specified.

  As a method of estimating the inclination of the acceleration sensor in a motion state of a fixed posture, there is a method of deriving the inclination of the acceleration sensor in a stationary state as the inclination of the motion state, but in this embodiment, it is obtained from the acceleration sensor in the motion state A method for estimating the inclination of the acceleration sensor in the motion state based on the acceleration data to be performed will be described. Although the posture of the acceleration sensor carried by a pedestrian changes every moment, the posture change within a short time of several seconds is not so large. Therefore, the posture of the acceleration sensor is considered to be constant in a short time interval during walking. Linear walking on a horizontal plane can be regarded as constant velocity linear motion. A curved walk on uneven ground can also be regarded as a constant-velocity linear motion in a time interval in which only a few steps are made, but not less than one step. The acceleration detected by the acceleration sensor in the constant-velocity linear motion state is an acceleration opposite to the gravitational acceleration. Therefore, an appropriate section is set, and the average of the acceleration data in the section is regarded as the acceleration opposite to the gravitational acceleration. Thus, when the acceleration regarded as the acceleration opposite to the gravitational acceleration is referred to as still data, the direction of the still data in the coordinate system unique to the acceleration sensor is estimated as being vertically upward. In this embodiment, the vertical direction is estimated based on the orientation of the acceleration sensor as described above, and this direction is derived as the inclination of the acceleration sensor. In this way, the axis that is regarded as the walking direction axis when deriving the inclination of the acceleration sensor is the horizontal vibration axis.

  So far, the principle of deriving the horizontal vibration axis regarded as the walking direction axis has been described. The direction to be estimated as the walking direction is any one of the two directions parallel to the horizontal vibration axis, but the walking direction cannot be estimated even if the spatial distribution of the walking data is analyzed. Therefore, acceleration data is treated as time-series information in order to determine one of the two directions parallel to the horizontal vibration axis as the walking direction.

  FIG. 2 is a graph in which the magnitude of the horizontal vibration axis component and the magnitude of the vertical axis component of acceleration data output from the acceleration sensor from a stationary state to a walking state are plotted in time series. FIG. 2 plots experimental values obtained by subtracting stationary data from acceleration data so that the acceleration is 0 in a stationary state. When the acceleration data is offset so as to indicate 0 in the stationary state in this way, the offset acceleration data becomes data representing the acceleration detected by the acceleration sensor.

  As shown in FIG. 2, in the walking state, the vibration phase of the horizontal vibration axis component of the acceleration data advances a little compared to the vibration phase of the vertical axis component. Therefore, by determining the positive / negative direction of the horizontal vibration axis of the acceleration data so that the vibration phase of the horizontal vibration axis component of the acceleration data advances rather than being delayed with respect to the vertical axis component, either of the two directions parallel to the horizontal vibration axis can be determined. Whether it is a walking direction can be determined.

(2) Hardware Configuration of Walking Navigation System FIG. 3 is a block diagram showing a hardware configuration of the walking navigation system as one embodiment of the present invention. FIG. 3 shows a main part of a walking navigation system configured as various portable information terminals such as a mobile phone terminal, a PDA, and a portable navigation device.

  The acceleration sensor 10 is a three-dimensional acceleration sensor that detects acceleration by decomposing into three orthogonal axes. The acceleration sensor 10 may be any detection method such as a piezoresistive type, a capacitance type, or a heat detection type. The acceleration sensor 10 outputs acceleration data g (x, y, z) representing acceleration obtained by combining the acceleration opposite to the gravitational acceleration and the acceleration inherent to the motion of the acceleration sensor. That is, for example, the acceleration data output from the acceleration sensor 10 is (0, 0, 0) when the acceleration sensor 10 is naturally falling in a posture in which the positive direction of the z-axis is vertically upward, and is stationary in the same posture. In this state, the acceleration data output from the acceleration sensor 10 is (0, 0, g) (g is a positive constant), and in the same posture, the acceleration is vertically upward and the absolute value of acceleration is equal to the absolute value of gravity acceleration. The acceleration data output from the acceleration sensor 10 in the state of linear motion is (0, 0, 2g).

  The magnetic sensor 11 includes an MI element, an MR element, and the like, and is a sensor that detects magnetism by decomposing into three orthogonal axes. The direction of each axis of the xyz coordinate system unique to the magnetic sensor 11 matches the direction of each axis of the xyz coordinate system unique to the acceleration sensor 10. That is, the magnetic data h (x, y, z) = (1, 1, 1) output from the magnetic sensor 11 and the acceleration data g (x, y, z) = (1, 1) output from the acceleration sensor 10. The magnetic sensor 11 and the acceleration sensor 10 are fixed to a common substrate 17, for example, so that 1) faces in the same direction.

  A GPS (Global Positioning System) unit 12 is an antenna for receiving orbit data transmitted from three or four satellites used for satellite navigation, and an ASIC (Application Specific Integrated) for outputting latitude and longitude data indicating the current position. Circuit).

The PU (Processor Unit) 14 outputs a walking direction and controls various hardware of the walking navigation system by executing a walking navigation program, an OS (Operating System), and the like.
A ROM (Read Only Memory) 15 is a non-volatile storage medium in which various computer programs executed by the PU 14 are stored.
A RAM (Random Access Memory) 16 is a storage medium that functions as a work memory for the PU 14 to execute a computer program and store acceleration data.
The IF (Inter Face) 13 includes an AD converter, a DA converter, and the like, and sets the format of data transferred between the acceleration sensor 10, the magnetic sensor 11, the GPS unit 12, the PU 14, the ROM 15, the RAM 16, and the like on the receiving side. This module converts data format.

(3) Data Flow of Walking Navigation System FIG. 4 is a data flow diagram showing the main part of the data flow of the walking navigation system. The process shown in FIG. 4 is processed by the PU 14 that executes the walking navigation program.

  The inclination estimation process 21 receives acceleration data g (x, y, z), and derives (α, β) as an estimated value of the inclination of the acceleration sensor 10 based on stationary data that is an average of acceleration data. The estimated value of the inclination of the acceleration sensor 10 is the direction of the static data defined in the xyz coordinate system unique to the acceleration sensor 10, and the angle α (y between the x axis unique to the acceleration sensor 10 and the xy component of the static data The positive direction of the axis is −180 ° <α ≦ 180 °) and the angle β formed by the z-axis inherent to the acceleration sensor 10 and the stationary data (−90 ° <+ with the positive direction of the z-axis being positive) β ≦ 90 °).

  The rotation process 20 receives the acceleration data g (x, y, z) and the estimated value (α, β) of the inclination of the acceleration sensor 10 and rotates the coordinate axes of the acceleration data based on these values, thereby The acceleration data is converted into data represented by an XYZ orthogonal coordinate system in which the stationary data that the acceleration sensor detects is taken in the positive direction of the Z axis, and as a result, acceleration data g (X, Y, Z) is output.

  The horizontal vibration axis deriving process 22 inputs the horizontal component g (X, Y) of acceleration, which is the XY component of the acceleration data g (X, Y, Z), for an arbitrary section, and analyzes the distribution to thereby obtain the horizontal vibration axis. The slope of is derived. The inclination of the horizontal vibration axis is derived as an angle θ (−90 ° <θ ≦ 90 °) formed with the X axis of the XYZ orthogonal coordinate system. The arbitrary section is a section serving as a unit for deriving one walking direction in the autonomous navigation section (see FIG. 5) in which the GPS unit 12 cannot receive the trajectory data. The arbitrary section may be divided by time or by the number of steps.

  The horizontal vibration axis component derivation process 30 inputs the horizontal component g (X, Y) of acceleration and the inclination θ of the horizontal vibration axis, and rotates the coordinate axis of the acceleration data based on these to convert the horizontal vibration axis to t. Acceleration data is converted into data represented by a tu orthogonal coordinate system having an axis and an axis perpendicular to the t-axis and included in the horizontal plane as the u-axis, and as a result, horizontal vibration axis component data g (t) of acceleration is derived. .

  The phase difference determination process 23 inputs g (t) which is a horizontal vibration axis component of acceleration and a vertical axis component g (Z) of acceleration which is a Z component of acceleration data g (X, Y, Z), and accelerates the acceleration. It is determined whether or not the phase of the horizontal vibration axis component of the horizontal component is advanced with respect to the phase of the vertical axis component of acceleration.

  The walking direction derivation process 26 inputs θ which is the inclination of the horizontal vibration axis and true / false which is the determination result of the phase difference determination process 23, and derives the walking direction. The walking direction is θ (−90 ° <θ ≦ 90 °) or θ + 180 °, which is an angle formed by the X axis of the XYZ orthogonal coordinate system and the walking direction.

  The walking direction deriving process 28 inputs the walking direction, the magnetic data h (x, y, z), and the inclination (α, β) of the acceleration sensor 10, and based on these, the north direction and the walking direction are formed. The walking direction which is an angle is derived.

  The above process is a process for deriving the walking direction of an arbitrary section. Next, a process for deriving the walking distance of an arbitrary section will be described.

  The step deriving process 24 inputs the vertical axis component g (Z) of the acceleration data for the step length measurement section and the arbitrary section from the step calculation place to the step settlement position, and based on this, the step number n of the arbitrary section and the step calculation are calculated. The number of steps N from the ground to the number of steps settlement is derived. The stride measurement section is a section in which a reference stride that is an average stride of a pedestrian is derived in a satellite navigation section (see FIG. 5) in which the GPS unit 12 can receive orbit data. The step length measurement section may be divided by time or by the number of steps. As a method for deriving the number of steps n in an arbitrary section, a method is known in which the number of times that the acceleration data becomes 0 in the arbitrary section is counted and the number of steps n is derived based on this. There is a difference in the time when the magnitude of the walking direction axis component and the vertical axis component of the acceleration of the acceleration sensor 10 is 0, and it is difficult to set an appropriate threshold value that considers the magnitude of the component to be 0. There is. For this reason, it is desirable to derive the number of steps n using only the vertical axis component of the acceleration data.

  The stride derivation process 25 inputs the step count N of the stride measurement section and the GPS position data p (x, y) of the step count calculation location and the step count settlement location, and derives the reference stride w based on these.

  The walking distance deriving process 27 inputs the number of steps n in an arbitrary section and the reference stride w and multiplies them to derive the walking distance d.

  The above is the process for deriving the walking distance of an arbitrary section. Next, a current position deriving process 29 for deriving the current position of the pedestrian will be described.

The current position deriving process 29 inputs the walking direction of the arbitrary section, the walking distance of the arbitrary section, and the GPS position data of the satellite navigation section, and derives the current position P (x, y) based on these. The current position P (x, y) is vector data indicating the latitude and longitude of the current location. The current position deriving process 29 outputs the GPS position data as the current position as it is in the satellite navigation section shown in FIG. 5, and derives the current position as follows in the autonomous navigation section. That is, when the GPS unit 12 cannot receive the orbit data, the GPS position data acquired last is set as the start point of the first arbitrary section. Then, walking vector of any section to be derived based on the arbitrary section _{(p 3 / p 4, p} 4 / p 5, p 5 / p 6, p 6 / p 7, ...) walking azimuth and distance walked Is added to the starting point of the first arbitrary section to derive the current position P (x, y) of the autonomous navigation section.

(4) Details of Method for Deriving Walking Direction Next, details of a method for deriving a walking direction by the above-described navigation system will be described.
FIG. 6 is a diagram in which acceleration data of an arbitrary section during walking is plotted in an xyz coordinate system unique to the acceleration sensor 10. As shown in FIG. 6, the acceleration data during walking is distributed in a finite range in which the central portion is separated from the origin. The reason why the central portion of the distribution of the acceleration data is separated from the origin is that the acceleration sensor 10 outputs data representing the acceleration obtained by combining the acceleration opposite to the gravitational acceleration and the acceleration inherent to the walking motion. As described above, in the present embodiment, the average of acceleration data is regarded as acceleration in the direction opposite to the gravitational acceleration and is defined as stationary data, and the direction of stationary data defined in the coordinate system unique to the acceleration sensor 10 is defined as the inclination of the acceleration sensor. To derive. FIG. 7 is a table showing a comparison result between the average acceleration data for the section where the user walks for 10 seconds from the stationary state and the acceleration data output from the acceleration sensor 10 in the stationary state at the start of walking. From the comparison results shown in FIG. 7, it can be seen that the inclination of the acceleration sensor can be estimated within a practically reliable range even if the average of the acceleration data is derived as stationary data. For example, even if the average acceleration data is derived as stationary data, it is possible to derive the walking direction with an error range of about ± 5 °.

Specifically, the inclination of the acceleration sensor 10 is estimated as follows, for example.
_First, stationary data g _ave (x, y, z) that is an average of acceleration data g (x, y, z) is derived for a section for estimating the inclination of the acceleration sensor 10. This section may or may not coincide with an arbitrary section that derives the walking direction using the estimated tilt value of the acceleration sensor 10, may be a continuous section, or may be separated from each other. A plurality of sections may be used. For example, the stationary data may be derived in a section immediately before the section from which the walking direction is derived. In addition, the direction of still data derived in a certain section may be used for deriving the walking direction in a plurality of sections after the section.

Next, the inclination of the acceleration sensor 10 is specified in the xyz coordinate system which is a coordinate system unique to the acceleration sensor 10. Specifically, the direction of still data is defined by two values α and β, and the inclination of the acceleration sensor 10 is specified as (α, β). α is an angle formed by the xy component of the still data and the x-axis, and is derived as an angle (°) of the angle formed by the vectors (x, y) and (1, 0) that are the xy component of the still data. α is positive in the direction from (0, 1) to (1, 0). β is an angle between the static data and the z-axis, and is the unit of the same coordinate plane as the static data ((x ² + y ² ) ^1/2 , z) parallel to the static data and expressed in the coordinate plane including the z-axis Derived as an angle with the vector (0, 1).

Derivation of horizontal component and vertical axis component of acceleration FIG. 8 is a diagram in which acceleration data is plotted in an XYZ orthogonal coordinate system in which the direction of still data is the positive direction of the Z axis. In this XYZ orthogonal coordinate system, it is estimated that the positive direction of the Z axis is vertically upward, and the XY plane is estimated to be parallel to the horizontal plane.

The mapping of the coordinate axis conversion from the xyz orthogonal coordinate system unique to the acceleration sensor 10 to the XYZ orthogonal coordinate system is expressed by the product of the matrix B of the following equation (4) and acceleration data g (x, y, z).

  Matrix representation of mapping from xyz coordinate system inherent to acceleration sensor 10 to XYZ coordinate system by calculating the product of the two matrices on the right side of equation (4) based on the inclination (α, β) of acceleration sensor 10 A matrix B is derived. In addition, there is a method of using a quaternion as another method of coordinate axis conversion from the xyz orthogonal coordinate system to the XYZ orthogonal coordinate system.

  Next, the coordinate axis of the acceleration data is converted using the matrix B representing the mapping from the xyz coordinate system unique to the acceleration sensor 10 to the XYZ coordinate system, and the XYZ coordinate system in which the direction of the stationary data is the positive direction of the Z axis. The represented acceleration data g (X, Y, Z) is derived.

Derivation of horizontal vibration axis FIG. 9 is a diagram in which acceleration data is projected on the XY plane of the XYZ orthogonal coordinate system. That is, the data indicated by the solid line in FIG. 9 is the XY component of the acceleration data, and is two-dimensional vector data regarded as the horizontal component of acceleration. 10 and 11 show the horizontal vibration axis that is the major axis of the distribution of the XY component that is the horizontal component of the acceleration data. FIG. 10 shows the horizontal vibration axis in the XYZ coordinate system in which the XY plane is estimated to be a horizontal plane, and FIG. 11 shows the horizontal vibration axis in the xyz coordinate system unique to the acceleration sensor 10. As shown in FIGS. 10 and 11, the horizontal component of acceleration during walking shows a flat distribution in which the lengths of the major axis and the minor axis are clearly different. As described above, in order to specify one of the two directions parallel to the horizontal vibration axis as the walking direction, the distribution of the horizontal component of acceleration is analyzed, and the angle θ formed by the horizontal vibration axis that is the major axis of the distribution and the X axis The horizontal vibration axis is derived by calculating (−90 ° <θ ≦ 90 °) by the following method.

The mapping of the coordinate axis conversion of the horizontal component g (X, Y) of the acceleration data to the tu orthogonal coordinate system with the horizontal vibration axis as the t-axis is expressed by the following equation (5).

When the t-axis is parallel to the horizontal vibration axis, the mapping of equation (5) corresponds to the principal axis conversion of the horizontal component distribution of the acceleration data. Since the dispersion becomes maximum in the major axis direction of the distribution, when the t-axis is parallel to the horizontal vibration axis, f (θ) represented by the following equation (6) becomes a maximum.

In Expression (6), an array Q of acceleration g (X, Y) in an arbitrary section from which a walking azimuth is derived is defined by the following Expression (7).

When f (θ) takes an extreme value, f ′ (θ) represented by the following equation (8) is zero.

Therefore, two θ ₁ and θ ₂ satisfying the following formula (9) are calculated and substituted for f (θ), respectively, and the angle at which f (θ) becomes larger among θ ₁ and θ ₂ is specified and specified. The calculated angle is derived as θ formed by the horizontal vibration axis and the X axis.

また、水平加速度の分散が最小となる分布の短軸の傾きを導出し、短軸の傾きから長軸の傾きを導出しても良いことはいうまでもない。 In order to prevent the calculation of Expression (9) from diverging, it is necessary to calculate in advance the absolute value of the denominator and the absolute value of the numerator on the right side of Expression (9), and to calculate using the larger absolute value as the denominator. There is. Therefore, the absolute value of the denominator and the absolute value of the numerator on the right side of Equation (9) are calculated, and when the absolute value of the numerator is smaller, θ is derived using Equation (9), and the absolute value of the numerator When is larger, θ is derived using the following equation (10) instead of equation (9).

It goes without saying that the inclination of the short axis of the distribution with the minimum horizontal acceleration variance may be derived, and the inclination of the long axis may be derived from the inclination of the short axis.

  When the processing so far is completed, the inclination of the horizontal vibration axis is derived. However, the horizontal vibration axis is an axis having no meaning in the positive and negative directions. In the processing so far, the inclination of the horizontal vibration axis having no meaning in the positive / negative direction is derived by treating the acceleration data as time series data in an arbitrary section as simple spatial information.

Derivation of walking direction FIG. 12 is a graph in which the horizontal vibration axis component t of acceleration data is plotted on the assumption that the horizontal vibration axis component t of acceleration data matches the walking direction axis component of acceleration data. Since the horizontal axis p is p used as a subscript of each element of the array Q of the horizontal acceleration g (X, Y), it represents the discrete time when the acceleration sensor 10 outputs the acceleration data. FIG. 13 is a graph showing the vibration of the walking direction axis component and the vertical axis component of the acceleration of the acceleration sensor 10 carried by the pedestrian (this acceleration is a true acceleration). When there is no measurement error in the acceleration data and there is no error in the estimated inclination of the acceleration sensor 10, the walking direction axis component of the true acceleration coincides with the horizontal vibration axis component t or -t of the acceleration data, and the true acceleration The vertical axis component coincides with the Z component regarded as the vertical component of the acceleration data. The measurement error of the acceleration data is sufficiently reduced by a known offset correction (the offset correction here is not correction for subtracting still data but processing for correcting temperature characteristics and the like). Further, as described above, the estimation accuracy of the inclination of the acceleration sensor 10 is sufficiently high. Therefore, the vibration phase of the horizontal vibration axis component t or −t of the acceleration data slightly advances with respect to the vibration phase of the Z component of the acceleration data. Therefore, the horizontal vibration axis component t and the vertical axis component Z of the acceleration data are treated as time series data, for example, by the following method, the vibration phase of the horizontal vibration axis component t of the acceleration data and the vibration phase of the vertical axis component Z of the acceleration data are Based on the result, one of the two directions parallel to the horizontal vibration axis is determined as the walking direction.

First, the stationary data is subtracted from the time point p _t [i] (i = 0, 1, 2,...) When the horizontal vibration axis component t of acceleration data becomes 0 from positive to negative and the vertical axis component Y of acceleration. _Deriving points p _z [j] (j = 0, 1, 2,...) When the sign of the value becomes 0 from positive to negative, respectively, and for each i, p _z [j] and p _z [j + 1] ] P _t [i] in between. Then, it is determined whether f (i, j) in the following equation (11) is larger or smaller than 0.5. If it is smaller, θ is the walking direction, and if larger, θ + 180 ° is the walking direction.

  The above-described determination method using equation (11) is that the difference φ between the vibration phase of the walking direction axis component and the vibration phase of the vertical axis component with respect to the acceleration of the acceleration sensor 10 (φ is the direction in which the vibration phase of the walking direction axis component advances) Is a value that is always less than ½ of the vibration period T of the walking direction axis component of the acceleration of the acceleration sensor 10. As for the acceleration of the acceleration sensor 10 carried by the pedestrian, the vibration period of the walking direction axis component and the vibration period of the vertical axis component substantially coincide with each other. You may make it correspond.

式（１２）を満たすｐが導出されたら、次にｐとｐ＋１との間においてｔ〔ｐ〕を直線近似して値が０となる時点に対応する水平振動軸成分ｐ_ｔ〔ｉ〕を導出する。具体的には次式（１３）を満たすｐ_ｔ〔ｉ〕を導出する。

ｐ_ｚ〔ｊ〕もｐ_ｔ〔ｉ〕と同様にして導出することができる。 The derivation of p _t [i] and p _z [j] is performed as follows, for example.
When the horizontal vibration axis component t and the vertical axis component Z of the acceleration data are respectively expressed as functions of p representing the discrete time when the acceleration sensor 10 outputs the acceleration data, t [p] and Z [p] are obtained. A time interval at which the acceleration sensor 10 outputs acceleration data (that is, a time interval at which acceleration data is read) is denoted by Δp.
First, in order to capture the timing when the horizontal vibration axis component of acceleration data changes from positive to negative, p satisfying the following equation (12) is derived.

When p satisfying Expression (12) is derived, t [p] is next linearly approximated between p and p + 1 to derive the horizontal vibration axis component p _t [i] corresponding to the time point when the value becomes 0. To do. Specifically, p _t [i] that satisfies the following equation (13) is derived.

p _z [j] can also be derived in the same manner as p _t [i].

ただしＣは、例えば０．１Ｇ相当の正の定数とする。
また加速度データの出力間隔が狭い場合には、１個とばし、２個とばしというようにとびとびの加速度データを式（１２）に代入して加速度データの水平振動軸成分が正から負に転ずるタイミングを捕捉してもよい。 However, if the robot is not completely stationary even when it is not walking, the timing that becomes noise is captured by Equation (12). Therefore, it is desirable to capture as a significant timing for deriving the vibration phase difference only when the following equation (14) is satisfied in addition to the equation (12).

However, C is a positive constant equivalent to 0.1 G, for example.
When the output interval of acceleration data is narrow, skipping one piece and skipping two pieces of acceleration data is substituted into equation (12), and the timing at which the horizontal vibration axis component of the acceleration data turns from positive to negative is set. May be captured.

  As another method for determining one of the two directions parallel to the horizontal vibration axis as the walking direction, f (i, j) in which the denominator on the left side of Equation (11) is replaced is compared with 2 in magnitude. Of course, the method may be determined by the sign of the integrated value of the difference between the horizontal vibration axis component t of the acceleration data and the value obtained by subtracting the stationary data from the vertical axis component Z of the acceleration data, A method is conceivable in which the determination is made based on the positive / negative of the Y component of the acceleration data when the horizontal vibration axis component t of the acceleration data becomes zero.

  Since θ is an angle formed with the X axis of the XYZ Cartesian coordinate system in which still data is taken positively on the Z axis, the walking direction derived as θ or θ + 180 ° by the above-described method is as shown in FIGS. The direction is based on the attitude of the acceleration sensor 10 regarded as constant.

-Derivation of walking direction The method of deriving the walking direction based on the four directions of east, west, south, and north from the walking direction based on the posture of the acceleration sensor 10 is as follows.
Under the precondition that the direction of each axis of the xyz coordinate system unique to the magnetic sensor 11 is the same as the direction of each axis of the xyz coordinate system unique to the acceleration sensor 10, the three-axis orthogonal coordinate system of the acceleration data g The rotation of the coordinate axis that matches the direction of each axis of the magnetic data h and the direction of each axis of the three-axis orthogonal coordinate system of the magnetic data h makes the north direction indicated by the magnetic data and the walking direction based on the attitude of the acceleration sensor 10 The angle D (°) is a walking direction with reference to east, west, south, and north (see FIG. 1).

Therefore, the coordinate system of the magnetic data is converted from the xyz orthogonal coordinate system unique to the magnetic sensor 11 to the XYZ orthogonal coordinate system using the mapping represented by the following equation (15), and the magnetic data (h _X , h _{Y of the} XYZ orthogonal coordinate system is converted. , H _Z ), the walking direction D with reference to the east, west, south, and north can be derived by subtracting the walking direction θ or θ + 180 ° from the azimuth angle indicated by h _Z ).

(5) Effects In the walking navigation method described above, the walking direction is derived by statistical calculation based on a plurality of acceleration data acquired from the three-dimensional acceleration sensor in an arbitrary section. For this reason, the walking direction can be derived with higher accuracy than the method of deriving the walking direction for each step by appropriately setting the length of the arbitrary section that divides the population. That is, in the method of deriving the walking direction for each step, since the walking direction is determined for each step, the determined error is accumulated every time the walking direction is determined. On the other hand, in the method of the above-described embodiment, for example, the error of the walking direction, the walking direction, and the current position is derived by statistically processing acceleration data divided by an arbitrary section that advances, for example, 5 steps and 10 steps. Get smaller.
In addition, since the average of acceleration data is derived as still data corresponding to the acceleration in the direction opposite to the gravitational acceleration, even if the posture of the acceleration sensor 10 changes during walking, the walking direction, walking direction, Position errors can be reduced during walking.
(6) Other Embodiments Although the present invention has been described above using the embodiments, it goes without saying that the technical scope of the present invention is not limited to the above-described embodiments, and does not depart from the gist of the present invention. Of course, various modifications can be made to the above-described embodiment.

It is a schematic diagram concerning the embodiment of the present invention. It is a graph which concerns on embodiment of this invention. It is a block diagram concerning the embodiment of the present invention. It is a data flow figure concerning the embodiment of the present invention. It is a schematic diagram concerning the embodiment of the present invention. It is a schematic diagram concerning the embodiment of the present invention. It is a table | surface which concerns on embodiment of this invention. It is a schematic diagram concerning the embodiment of the present invention. It is a schematic diagram concerning the embodiment of the present invention. It is a schematic diagram concerning the embodiment of the present invention. It is a schematic diagram concerning the embodiment of the present invention. It is a graph which concerns on embodiment of this invention. It is a graph which concerns on embodiment of this invention. It is a schematic diagram concerning the embodiment of the present invention. It is a schematic diagram concerning the embodiment of the present invention.

Explanation of symbols

10: acceleration sensor, 11: magnetic sensor, 12: GPS unit, 13: IF, 14: PU, 15: ROM, 16: RAM, 17: substrate, 20: rotation process, 21: tilt estimation process, 22: horizontal vibration Axis derivation process, 23: Phase difference determination process, 24: Step count derivation process, 25: Step length derivation process, 26: Walking direction derivation process, 27: Walking distance derivation process, 28: Walking direction derivation process, 29: Current position derivation process , D: angle (walking direction), d: walking distance

Claims (9)

Acquiring the acceleration data at a discrete time within an arbitrary interval corresponding to a plurality of steps from a three-dimensional acceleration sensor that outputs data indicating a combined vector of gravity and inertial force acting on the unit mass as acceleration data;
Deriving a horizontal component and a vertical component of the acceleration of the three-dimensional acceleration sensor based on the acceleration data,
The direction in which the variance of the horizontal component of the acceleration is maximized with respect to the attitude of the three-dimensional acceleration sensor is derived as the inclination of the horizontal vibration axis ,
Deriving a horizontal vibration axis component of the acceleration based on the inclination of the horizontal vibration axis,
Based on the deviation between the vibration phase of the horizontal vibration axis component of the acceleration and the vibration phase of the vertical axis component of the acceleration, one of the two directions parallel to the horizontal vibration axis is set to the tertiary in the arbitrary section. The walking direction is determined based on the posture of the original acceleration sensor.
Look at including it,
Deriving a mapping from the coordinate system unique to the three-dimensional acceleration sensor to an XYZ orthogonal coordinate system in which the average direction of the acceleration data in the time interval corresponding to a plurality of steps is parallel to the Z axis,
The horizontal component of the acceleration is an X, Y component of the acceleration data in the XYZ orthogonal coordinate system,
The vertical axis component of the acceleration is a Z component of the acceleration data in the XYZ orthogonal coordinate system.
Walking navigation method. The deviation between the vibration phase of the horizontal vibration axis component of the acceleration and the vibration phase of the vertical axis component of the acceleration is determined by the vibration phase of the vertical axis component of the acceleration with respect to the vibration phase of the horizontal vibration axis component of the acceleration. The ratio between the delay and the period of the horizontal vibration axis component of the acceleration or the vibration period of the vertical axis component of the acceleration,
The walking navigation method according to claim 1 . Obtain magnetic data from a magnetic sensor that changes posture with the three-dimensional acceleration sensor,
Deriving the walking direction in the arbitrary section based on the magnetic data and the walking direction,
Walking navigation method according to any one of claims 1 and 2. Deriving the number of steps in the arbitrary section based on the vertical axis component of the acceleration,
Deriving the walking distance of the arbitrary section based on the reference stride and the number of steps of the arbitrary section,
The walking navigation method according to any one of claims 1 to 3 . Deriving the section distance from the step count calculation location to the step count calculation location based on the GPS location data of the step count calculation location and the GPS location data of the step count settlement location,
Deriving a reference number of steps from the step count value to the step count settlement value based on the vertical axis component of the acceleration,
Deriving the reference stride based on the section distance and the reference step count;
The walking navigation method according to claim 4 . Deriving a current position based on the walking direction, the walking distance, and GPS position data,
The walking navigation method according to claim 4 or 5 . When the horizontal component of the acceleration in the arbitrary section is (X _p , Y _p ), where (p = 0, 1, 2,... S), θ (−90 ° < θ ≦ 90 °) is derived as the inclination of the horizontal vibration axis.
The walking navigation method according to any one of claims 1 to 6 .

Means for acquiring the acceleration data at a discrete time within an arbitrary interval corresponding to a plurality of steps from a three-dimensional acceleration sensor that outputs data indicating a combined vector of gravity and inertial force acting on a unit mass as acceleration data;
Means for deriving a horizontal component and a vertical component of the acceleration of the three-dimensional acceleration sensor based on the acceleration data;
Means for deriving, as an inclination of a horizontal vibration axis, a direction in which a variance of the horizontal component of the acceleration is maximized with reference to an attitude of the three-dimensional acceleration sensor ;
Based on the deviation between the vibration phase of the horizontal vibration axis component of the acceleration and the vibration phase of the vertical axis component of the acceleration, one of the two directions parallel to the horizontal vibration axis is set to the tertiary in the arbitrary section. Means for determining a walking direction based on the posture of the original acceleration sensor;
With
Deriving a mapping from the coordinate system unique to the three-dimensional acceleration sensor to an XYZ orthogonal coordinate system in which the average direction of the acceleration data in the time interval corresponding to a plurality of steps is parallel to the Z axis,
The horizontal component of the acceleration is an X, Y component of the acceleration data in the XYZ orthogonal coordinate system,
The vertical axis component of the acceleration is a Z component of the acceleration data in the XYZ orthogonal coordinate system.
Walking navigation system. Means for acquiring the acceleration data at a discrete time within an arbitrary section corresponding to a plurality of steps from a three-dimensional acceleration sensor that outputs data indicating a combined vector of gravity and inertial force acting on the unit mass as acceleration data;
Means for deriving a horizontal component and a vertical component of the acceleration of the three-dimensional acceleration sensor based on the acceleration data;
Means for deriving, as an inclination of a horizontal vibration axis, a direction in which a variance of the horizontal component of the acceleration is maximized with reference to an attitude of the three-dimensional acceleration sensor ;
Based on the deviation between the vibration phase of the horizontal vibration axis component of the acceleration and the vibration phase of the vertical axis component of the acceleration, one of the two directions parallel to the horizontal vibration axis is set to the tertiary in the arbitrary section. Means for determining a walking direction based on the posture of the original acceleration sensor;
Function as a computer
Deriving a mapping from the coordinate system unique to the three-dimensional acceleration sensor to an XYZ orthogonal coordinate system in which the average direction of the acceleration data in the time interval corresponding to a plurality of steps is parallel to the Z axis,
The horizontal component of the acceleration is an X, Y component of the acceleration data in the XYZ orthogonal coordinate system,
The vertical axis component of the acceleration is a Z component of the acceleration data in the XYZ orthogonal coordinate system.
A walking navigation program.

Images