JP2013132968A - Vehicle attitude estimating device and theft informing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To highly accurately estimate a vehicle attitude, by correcting an offset error of an acceleration sensor with a simpler constitution.SOLUTION: An acceleration sensor detects acceleration generated in a vehicle, and a GPS receiver outputs GPS data based on GPS positioning data. An attitude estimating part calculates an observation value Y (S320), calculate a system matrix φ (S330), makes a prediction calculation of an error covariance value P (S340), calculates Kalman gain K (S350), and calculates the error covariance value (S360), Thus, an offset error is estimated as a state quantity x (S370), and the offset error of the acceleration sensor is updated (S380). The attitude estimating part arithmetically operates a vehicle attitude vector by using the offset error.

Description

  The present invention relates to a vehicle posture estimation device that estimates the posture of a vehicle, and a theft notification device that uses the vehicle posture estimation device.

  A stolen vehicle tracking service (hereinafter referred to as “SVT”) that detects a vehicle theft accident in a device mounted on a vehicle and notifies the security center and the person of the location information of the stolen vehicle using positioning means such as GPS. ") Is known.

  One technique for detecting such a theft accident is to detect a change in the attitude of the vehicle. The change in the posture of the vehicle is detected using, for example, a three-axis acceleration sensor. Specifically, when the vehicle is stopped, the acceleration of gravity is detected by the acceleration sensor. Therefore, it is possible to detect a change in the attitude of the vehicle from this change in gravitational acceleration, and to detect a theft accident.

  However, it is known that the acceleration sensor includes an error in the measurement value. This error is roughly classified into a sensor mounting angle error and an offset error depending on a temperature change. Since the measurement value of the acceleration sensor includes an error in this way, it is necessary to accurately estimate the error of the acceleration sensor in order to accurately detect the change in the posture of the vehicle.

  Conventionally, a method of estimating an error of a triaxial acceleration sensor using vehicle speed information via CAN has been disclosed (for example, see Patent Document 1). In addition, a method for estimating the mounting angle error of the biaxial and triaxial acceleration sensors using a vehicle speed or a gyroscope is disclosed (for example, see Patent Document 2). Furthermore, a technique for estimating an offset error of a biaxial acceleration sensor has been disclosed (see, for example, Patent Document 3).

US Patent Application Publication No. 2010/0318257 JP 2005-147696 A JP 2000-356647 A

  However, in the inventions described in Patent Documents 1 and 2, vehicle speed information is acquired via, for example, CAN, but there is a problem that the accuracy of error estimation depends on the accuracy of the vehicle speed information and the reception cycle. In addition, it is necessary to configure the system so that vehicle speed information can be acquired via CAN, and the system configuration may be complicated.

  On the other hand, in the invention described in Patent Document 3, it is necessary to accurately attach the attachment angle error to be close to “0” as a premise for estimating the offset error, which may result in poor versatility.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to correct the offset error of the acceleration sensor with a simpler configuration and estimate the vehicle posture with high accuracy.

  In the vehicle posture estimation apparatus according to claim 1, which is made to achieve the above object, the acceleration sensor detects acceleration generated in the vehicle. The GPS receiver outputs GPS data based on the GPS positioning data. The GPS data may be the GPS positioning data itself, or the current position of the vehicle and the vehicle speed obtained from the GPS positioning data.

  Here, in particular, the observation value calculation means calculates the observation value from the sensor data based on the acceleration obtained by the acceleration sensor and the GPS data. Specifically, the sensor data may be a vehicle speed obtained by integrating acceleration detected by the acceleration sensor.

  At this time, the error estimation means uses the offset value of the acceleration sensor as a state quantity, uses the observation value calculated by the observation value calculation means, and estimates the offset error by the Kalman filter. Then, the vehicle posture is estimated by the vehicle posture estimation means using the estimated offset error. The vehicle posture is calculated as a vehicle posture vector, for example.

  That is, in the present invention, the Kalman filter is used in order to combine with the GPS system. In this way, it is not necessary to acquire vehicle speed information via, for example, CAN, the offset error of the acceleration sensor can be corrected with a simpler configuration, and the vehicle attitude can be estimated with high accuracy.

  By the way, as shown in claim 2, it is conceivable that the error estimation means calculates the acceleration sensor mounting angle error based on the gravitational acceleration when the vehicle is stopped and the acceleration in the vehicle traveling direction when the vehicle is traveling. In this case, the vehicle posture estimation means estimates the vehicle posture using the attachment angle error and the offset error. In this way, the mounting angle error of the acceleration sensor is also corrected.

  From the viewpoint of vehicle attitude estimation for the prevention of theft accidents, as shown in claim 3, a configuration in which the error estimation means stores the offset error in a storage device is conceivable. At this time, the vehicle posture estimation means estimates the vehicle posture using the offset error stored in the storage device even after the power supply to the vehicle is stopped. After the power supply to the vehicle is stopped refers to a state in which the power supply to the vehicle is stopped by turning off the ignition switch, for example. In this way, it is possible to estimate the vehicle posture using the offset error estimated during vehicle travel.

  The offset error of the acceleration sensor depends on the temperature around the sensor. Therefore, when the temperature rises due to the start of power supply to the vehicle, the offset error increases. However, it is known that the offset error becomes almost stable after a certain period of time.

  Therefore, as shown in claim 4, it is conceivable that the error estimation means stores a plurality of offset errors in the storage means based on the elapsed time from the start of power supply to the vehicle. For example, when the offset error fluctuates before a certain time elapses, the offset error is stored as a fluctuating error, and when the offset error stabilizes after a certain time elapses, the offset error is stored as a stable error. In this case, the vehicle posture estimation means estimates the vehicle posture by selectively using a plurality of offset errors based on the elapsed time since the power supply to the vehicle is stopped. This makes it possible to estimate the vehicle posture with high accuracy even after the power supply to the vehicle is stopped.

By the way, if the vehicle speed is low, the reliability of the vehicle speed based on the GPS positioning data is lowered. Therefore, as shown in claim 5, it is preferable that the error estimation means estimates the offset error when the vehicle speed based on the GPS positioning data exceeds a predetermined speed. In this way, it is possible to estimate the offset error with higher accuracy.

  Note that as the error covariance value increases, the reliability of the estimated offset error decreases. Therefore, as shown in claim 6, it is conceivable that the error estimation means stores the offset error in the storage device when the error covariance value is below a predetermined threshold value. In this way, the reliability of the offset error stored in the storage device can be increased.

  Further, when the GPS positioning data is not output for a long time, a divergence occurs between the error covariance value and the actual error. That is, there is a phenomenon in which the error recognized by the Kalman filter is reduced even though the actual error is increased.

  Therefore, as shown in claim 7, the error estimation means is configured such that when the GPS positioning data cannot be obtained, or when the vehicle speed based on the GPS positioning data is equal to or lower than the predetermined speed, a predetermined time has elapsed from the previous estimation of the offset error. When the time has elapsed, it is preferable to recalculate the error covariance value. In this way, since the offset error is stored in the storage device based on the newly calculated error covariance value, the reliability of the offset error can be increased.

  Although the above has been described as the invention of the vehicle posture estimation device, as shown in claim 8, it can also be realized as a theft notification device comprising the vehicle posture estimation device and a theft determination notification means. At this time, the theft determination notification means determines the presence / absence of a theft accident based on the vehicle posture estimated by the vehicle posture estimation means of the vehicle posture estimation device and notifies the outside. In this way, the effect of preventing a theft accident based on the vehicle posture estimation is conspicuous.

It is a block diagram which shows schematic structure of a theft notification system. (A) is explanatory drawing which shows the definition of each axis | shaft of an acceleration sensor, (b) is explanatory drawing which shows the attachment angle error of the acceleration sensor of a yaw direction. (A) is explanatory drawing which shows the attachment angle error of the acceleration sensor of a pitch direction, (b) is explanatory drawing which shows the attachment angle error of the acceleration sensor of a roll direction. It is a flowchart which shows the vehicle attitude | position estimation process which an attitude | position estimation part performs. It is a schematic diagram which shows the time change of the offset error of an acceleration sensor. It is a flowchart which shows a vehicle attitude | position calculation process. It is a flowchart which shows the composite process with GPS. It is a flowchart which shows the observation value calculation process in a composite process with GPS. (A) is explanatory drawing which shows the ENU coordinate system of WGS84, (b) is explanatory drawing which shows the ECEF coordinate system of GWS84. It is a flowchart which shows an error memory process.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram illustrating a schematic configuration of a theft notification system according to the embodiment.
The theft notification system includes a theft notification device 10 and a wireless communication service provider (Mobile Network
Operator (hereinafter referred to as “MNO”) 50 and a service providing provider (Telematics Service Provider, hereinafter referred to as “TSP”) 60.

  The theft notification device 10 determines whether or not a vehicle theft accident has occurred, and if a theft accident has occurred, notifies the MNO 50 of the position of the vehicle wirelessly. The MNO 50 notifies the TSP 60 of the position of the vehicle, and the TSP 60 notifies the contractor and the like that a vehicle theft accident has occurred and the position of the vehicle.

Here, the theft notification device 10 includes a vehicle attitude estimation device 20, an SVT determination unit 30, and a radio unit 40.
The vehicle posture estimation device 20 outputs the vehicle posture to the SVT determination unit 30 as a vehicle posture vector. Thereby, the SVT determination unit 30 determines whether or not a theft accident has occurred from the change in the vehicle attitude vector, and if it is determined that the theft accident has occurred, the SVT determination unit 30 requests the radio unit 40 to make a call to the MNO 50. The wireless unit 40 is embodied as a wireless device using, for example, W-CDMA, wireless LAN, or the like.

  As a more specific configuration, the vehicle attitude estimation device 20 includes a GPS antenna 21, a GPS receiver 22, a triaxial acceleration sensor (hereinafter simply referred to as “acceleration sensor”) 23, an attitude estimation unit 24, a real-time clock (hereinafter “RTC”). 25) and a storage device 26.

The GPS antenna 21 is configured to receive radio waves transmitted from the GPS satellite 70.
The GPS receiver 22 obtains a vehicle position and velocity vector from GPS positioning data based on radio waves from the GPS satellite 70 received via the GPS antenna 21. The vehicle position and speed vector are input to the posture estimation unit 24. The vehicle position is also input to the SVT determination unit 30.

The acceleration sensor 23 detects the acceleration of the vehicle. The vehicle acceleration detected by the acceleration sensor 23 is input to the posture estimation unit 24.
In addition, the RTC 25 has a timekeeping function, and the time output from the RTC 25 is also input to the attitude estimation unit 24.

  The posture estimation unit 24 estimates an error of the acceleration sensor 23 and stores the error in the storage device 26 as a correction value. As a result, the posture estimation unit 24 reads the correction value from the storage device 26 even when the vehicle is stopped when the GPS receiver 22 is not functioning, that is, when the ignition switch is turned off. The vehicle posture is estimated by correcting the acceleration. At this time, a vehicle posture estimation vector is output. All the other blocks function even when the vehicle stops when the GPS receiver 22 does not function.

First, the acceleration detected by the acceleration sensor 23 is mathematically expressed.
FIG. 2A is an explanatory diagram showing the definition of each axis of the acceleration sensor 23. Here, the rotation direction with respect to the Z axis is the yaw direction, the rotation direction with respect to the Y axis is the pitch direction, and the rotation direction with respect to the X axis is the roll direction.

  At this time, as shown in FIG. 2B, the mounting angle error of the acceleration sensor 23 in the yaw direction is ψ. Similarly, θ is an attachment angle error in the pitch direction as shown in FIG. 3A, and φ is an attachment angle error in the roll direction as shown in FIG.

  Due to such an attachment angle error, the acceleration detected by the acceleration sensor 23 changes. When this is represented by a rotation matrix, the rotation matrix in the yaw direction is expressed by the following equation 1-1.

  The rotation matrix in the pitch direction is expressed by the following expression 1-2.

  Furthermore, the rotation matrix in the roll direction is expressed by the following Expression 1-3.

Therefore, when the true acceleration generated in the vehicle is α _in and the acceleration detected by the acceleration sensor 23 is α _out , the following equations 1-4 are derived.

  When the expression 1-4 is expanded, the following expression 1-5 is obtained.

The mounting angle error is obtained using the above-described expression 1-5.
First, consider the stopping condition. In the stopped state, only gravitational acceleration is detected. The gravitational acceleration is expressed by the following equation 2-1.

  Substituting this into the above expression 1-5 yields the following expression 2-2.

From this equation 2-2, the mounting angle error θ in the pitch direction and the mounting angle error φ in the roll direction are obtained.
Next, consider a straight-ahead state. In the straight traveling state, acceleration in the straight traveling direction and gravitational acceleration are detected. The acceleration in the straight direction is Vα Then, the acceleration α _in of the vehicle is expressed by the following formula 2-3.

  Therefore, the acceleration to be detected is obtained as Equation 2-4. At this time, since the mounting angle errors θ and φ are known, the mounting angle error ψ is obtained. These attachment angle errors are obtained, for example, by the posture estimation unit 24 and stored in the storage device 26.

The calculation of the X direction offset error correction amount, the Y direction offset error correction amount, and the Z direction offset error correction amount will be described later.
Next, vehicle posture estimation processing will be described. FIG. 4 is a flowchart showing a vehicle posture estimation process. This vehicle posture estimation process is repeatedly executed by the posture estimation unit 24 in FIG.

In the first S10, acceleration is acquired. In this process, the acceleration detected by the acceleration sensor 23 is acquired.
In subsequent S20, vehicle posture calculation processing is executed. In the vehicle attitude calculation process, the offset error of the acceleration sensor 23 is selectively used.

Here, the offset error of the acceleration sensor 23 will be described. FIG. 5 is a schematic diagram showing a change over time in the offset error of the acceleration sensor 23. The offset error varies depending on the temperature. Therefore, as shown in FIG. 5, the offset error increases with the passage of time since the supply of power to the vehicle is started by turning on the ignition switch. When the predetermined time T1 elapses, the offset error is almost stabilized. Therefore, in the present embodiment, two errors are stored: an error before the time T1 elapses (hereinafter referred to as “error at fluctuation”) and an error after the time T1 elapses (hereinafter referred to as “error at stability”). Keep it. In the vehicle attitude calculation process, one of the errors is selectively used as an offset error depending on the elapsed time from when the power supply to the vehicle is stopped by turning off the ignition switch.

Specifically, the process of S20 is as shown in FIG.
In the first S200, it is determined whether or not an elapsed time Toff from the time of stopping the power supply exceeds T2. This elapsed time Toff is calculated from the time inputted from the RTC 25 in FIG. If it is determined that Toff> T2 is satisfied (S200: YES), the vehicle is cooled as time elapses exceeding the time T2, so the error at the time of variation is applied in S210, and the process proceeds to S230. On the other hand, when it is determined that Toff ≦ T2 (S200: NO), an error at the time of stabilization is applied in S220, and the process proceeds to S230.

  In S230, a vehicle attitude vector is calculated. In this process, the vehicle attitude vector is calculated using the above equation 2-4. In the state where the offset error is not calculated, such as immediately after shipment from the factory, the initial value “0” of the offset error is applied.

  Returning to the description of FIG. 4, in S30, the combined processing with GPS is executed. FIG. 7 is a flowchart specifically showing the combined processing with GPS. In this embodiment, a Kalman filter is used for the combined processing with GPS.

  In first S300, it is determined whether there is GPS positioning data. This process determines whether or not valid positioning data has been obtained by the GPS receiver 22. If it is determined that there is GPS positioning data (S300: YES), the process proceeds to S310. On the other hand, when it is determined that there is no GPS positioning data (S300: NO), the process proceeds to S390. The case where valid positioning data cannot be obtained means that the GPS receiver 22 is not functioning, for example, when the radio wave from the GPS satellite 70 cannot be received in a tunnel or the like, or the ignition switch is turned off. There are cases.

  In S310, it is determined whether or not the vehicle speed obtained from the GPS positioning data (hereinafter referred to as “GPS speed”) exceeds a threshold value V1. When the vehicle speed is low (including a stop state), an accurate GPS speed cannot be obtained. Therefore, an offset error is estimated when the GPS speed exceeds the threshold value V1. Here, when GPS speed> V1 (S310: YES), the process proceeds to S320. On the other hand, when GPS speed ≦ V1 (S310: NO), the process proceeds to S390.

  In S320, the observed value Y is calculated. In this process, the difference between the velocity vector of the GPS velocity and the velocity vector based on the acceleration acquired from the acceleration sensor 23 is calculated. The observation value calculation process is as shown in FIG.

In first S321, vehicle acceleration is extracted. This process acquires the acceleration obtained by the acceleration sensor 23. The acceleration obtained here is the acceleration α _local in the local coordinate system, and at this time, each component is represented by α _local x, α _local y, and α _local z.

In subsequent S322, integration is executed. Thereby, the velocity vector Vα _local is obtained from the acceleration α _local of the local coordinate system. At this time, each component is expressed as Vα _local x, Vα _local y, and Vα _local z.

In the next step S323, the absolute azimuth obtained from the GPS positioning data of the GPS receiver 22 is used to convert the velocity vector into the ENU coordinate system. FIG. 9A shows an ENU coordinate system of WGS84 (World Geodetic System 1984). WGS84 is a GPS reference coordinate system. In the ENU coordinate system, a point near the ground surface is the origin, the zenith direction (upward on the vertical line) is the positive direction of the Z axis, and the east direction perpendicular thereto is X Let the axis be the north direction. That is, here, conversion from the vehicle center coordinate system to the ENU coordinate system is performed. Let Vα _{enu be} the velocity vector of the ENU coordinate system obtained here.

  If the absolute azimuth obtained from the GPS positioning data is θ (−180 degrees <θ ≦ 180 degrees, clockwise), the conversion equation is as shown in the following equation 3-1.

In subsequent S324, the current position obtained from the GPS positioning data of the GPS receiver 22 is used to _convert the velocity vector Vα _enu in the ENU coordinate system into a velocity vector in the ECEF coordinate system. FIG. 10B shows an ECEF coordinate system of WGS84. The ECEF coordinate system is called the earth center / earth fixed rectangular coordinate system. Let Vα _{ecef be} the velocity vector of the ECEF coordinate system obtained here. Each component is expressed as Vα _ecef x, Vα _ecef y, Vα _ecef z. A specific rotation matrix is represented by the following expression 3-2, where the latitude of the current position obtained from GPS positioning data is lat (rad) and longitude lon (rad).

Accordingly, the inverse matrix of the rotation matrix of the above formula 3-2, by multiplying the velocity vector V.alpha _enu of ENU coordinate system, the velocity vector V.alpha _ECEF the ECEF coordinate system is obtained. It is as shown in the following Expression 3-3.

  In subsequent S325, the observed value Y is calculated. The observed value Y is calculated by the following equation 3-4.

Here, the offset error of the acceleration sensor 23 is εGb, and the offset error in each direction is εGbx, εGby, εGbz. ΕVx _gps , εVy _gps , and εVz _gps are observation noises.

In the Kalman filter, the offset error is estimated by the least squares by the following cyclic calculation using the observed value Y obtained here.
Returning to the description of FIG. 7, in S330, the system matrix φ is calculated. When the offset error εGb is used, the state equation is expressed by the following equations 4-1 to 4-3.

εGbx (t) = εGbx (t−1) + εx Equation 4-1
εGby (t) = εGby (t−1) + εy Equation 4-2
εGbz (t) = εGbz (t−1) + εz Equation 4-3

Here, εx, εy, and εz are system noises.

In the Kalman filter, an offset error to be estimated is a state quantity x. This state quantity x is updated by the following state equation. Here, φ is the system matrix and v is the system noise.

x (t) = φx (t−1) + v (t) Equation 4-4

On the other hand, the following expression 4-5 is obtained from the above expressions 4-1 to 4-3.

  That is, the system matrix φ is expressed by the following expression 4-6.

  In subsequent S340, an error covariance P is predicted and calculated. The error covariance P is predicted and calculated by the following equation 4-7.

W is system noise.
In the next S350, a Kalman gain K is calculated. The Kalman gain K is calculated by the following equation 4-8.

  In subsequent S360, the error covariance P is calculated. The error covariance P is calculated by the following equation 4-9.

  In the next S370, the state quantity x is calculated. The state quantity x is calculated by the following expression 4-10.

In the next S380, the error of the acceleration sensor 23 is updated. That is, it is as shown by the following equations 4-11 to 4-13.

X direction offset error correction amount (t) = X direction offset error correction amount (t−1) −εGbx (t) Equation 4-11
Y-direction offset error correction amount (t) = Y-direction offset error correction amount (t−1) −εGby (t) Equation 4-12
Z-direction offset error correction amount (t) = Z-direction offset error correction amount (t−1) −εGbz (t) Equation 4-13

After the process of S380 is completed, the combined process with GPS is terminated.

In S390, which is shifted when a negative determination is made in S300 or S310 in FIG. 7, it is determined whether or not a predetermined time has elapsed. This process determines a case where GPS positioning data cannot be obtained for a long time. When the GPS positioning data cannot be obtained for a long period of time, a divergence occurs between the error covariance value and the actual error. That is, there is a phenomenon in which the error recognized by the Kalman filter is reduced even though the actual error is increased. Therefore, a prediction calculation of the error covariance value P is performed by determining the passage of a predetermined time. If it is determined that the predetermined time has passed (S390: YES), the system matrix φ is calculated in S400, the error covariance value P is calculated in S410, and then GPS
The combined processing with is terminated. Note that the processing of S400 and S410 is the same as the processing of S330 and S340. On the other hand, as long as the predetermined time has not elapsed (S390: NO), the processing of S400 and S410 is not executed, and the combined processing with GPS is terminated.

Returning to the description of FIG. 4, in S40, an error storing process is executed. FIG. 10 is a flowchart showing the error storing process.
In the first S400, it is determined whether or not the error covariance value P is below the threshold value P1. If the error covariance value P is less than P1, it can be said that the estimated value is reliable. When P <P1 (S400: YES), the process proceeds to S410. On the other hand, when P ≧ P1 (S400: NO), the subsequent process is not executed and the error storing process is terminated.

  In S410, it is determined whether or not an elapsed time Ton from the start of power supply by turning on the ignition switch is less than a threshold value T1. When the elapsed time Ton from the power supply start time is less than T1, it is estimated that the offset error is in the fluctuation period (see FIG. 5). If Ton <T1 (S410: YES), the offset error is stored as a variation error in S420, and then the error storage process is terminated. On the other hand, if Ton ≧ T1 (S410: NO), an offset error is stored as an error at the time of stabilization in S430, and then the error storage process is terminated.

  As described above in detail, in the vehicle attitude estimation device 20 of the present embodiment, the acceleration sensor 23 detects the acceleration generated in the vehicle, and the GPS receiver 22 outputs GPS data based on the GPS positioning data. Then, the posture estimation unit 24 calculates the observed value Y (S320 in FIG. 7), calculates the system matrix φ (S330), performs prediction calculation of the error covariance value P (S340), and sets the Kalman gain K to The error covariance value P is calculated (S360). Thereby, the offset error is estimated as the state quantity x (S370), and the offset error of the acceleration sensor is updated (S380). The posture estimation unit 24 calculates a vehicle posture vector using the offset error (S230 in FIG. 6).

  That is, in this embodiment, a Kalman filter is used in order to combine with the GPS system. In this way, it is not necessary to acquire vehicle speed information via, for example, CAN, the offset error of the acceleration sensor 23 can be corrected with a simpler configuration, and the vehicle attitude can be estimated with high accuracy.

  Further, in the present embodiment, the mounting angle error of the acceleration sensor 23 is obtained using the above expression 1-5 and stored in the storage device 26. Thereby, the mounting angle error of the acceleration sensor 23 is also corrected.

  Furthermore, in this embodiment, the offset error of the acceleration sensor 23 is stored in the storage device 26, and the stored offset error is stored even after the power supply to the vehicle is stopped in the vehicle attitude calculation process shown in FIG. Read (S210, S220) and calculate the vehicle attitude vector (S230). As a result, the vehicle posture can be estimated using the offset error estimated when the vehicle travels.

Further, in the present embodiment, in the error storing process shown in FIG. 10, a plurality of offset errors are stored in the storage device 26 based on the elapsed time Ton after the power supply to the vehicle is started. Specifically, when the elapsed time Ton is less than the predetermined time T1, the offset error is stored as a variation error (S410: YES, S420), and when the elapsed time Ton is equal to or longer than the predetermined time T1, the offset error is stored as a stable error. Store (S410: NO, S420). In the vehicle attitude calculation process shown in FIG. 6, a plurality of offset errors stored in the storage device 26 are selectively applied based on the elapsed time Toff after the power supply to the vehicle is stopped.
Specifically, when the elapsed time Toff from the power supply stop point exceeds the predetermined time T2, a fluctuation error is applied (S200: YES, S210), and when the elapsed time Toff is equal to or shorter than the predetermined time T2. An error at the time of stability is applied (S200: NO, S220), and a vehicle attitude vector is calculated (S230). Thereby, even after the power supply to the vehicle is stopped, the vehicle posture can be estimated with high accuracy.

  Furthermore, in this embodiment, when the GPS speed exceeds a predetermined threshold value V1 (S310 in FIG. 7: YES), an offset error is estimated. Thereby, the offset error can be estimated with higher accuracy.

  In the present embodiment, when the error covariance value P is below the predetermined threshold value P1 (S400: YES in FIG. 10), the offset error is stored (S420, S430). Thereby, the reliability of the offset error stored in the storage device 26 can be increased.

  Note that if the GPS positioning data is not output for a long time, a deviation occurs between the error covariance value and the actual error. That is, there is a phenomenon in which the error recognized by the Kalman filter is reduced even though the actual error is increased.

  Therefore, in this embodiment, when there is no GPS positioning data (S300: NO in FIG. 7) or when valid positioning data cannot be obtained (S310: NO), when a predetermined time has elapsed (S390: YES). The system matrix is calculated (S400), and the prediction calculation of the error covariance value P is performed (S410). In this way, since the offset error is stored in the storage device 26 based on the newly calculated error covariance value P (S400: YES, S420, S430 in FIG. 10), the reliability of the offset error is increased. Can be made.

  Further, in the theft notification device 10 of the present embodiment, based on the vehicle attitude vector from the attitude estimation unit 24, the SVT determination unit 30 determines whether a theft accident has occurred from the change in the vehicle attitude vector, and the theft accident. Is determined to have occurred, the wireless unit 40 is requested to make a call to the MNO 50. As a result, the effect of preventing a theft accident based on the vehicle posture estimation is conspicuous.

Note that the theft notification device 10 in this embodiment corresponds to a “theft notification device” in the claims, and the vehicle posture estimation device 20 corresponds to a “vehicle posture estimation device”.
The acceleration sensor 23 in the present embodiment corresponds to an “acceleration sensor”, the GPS receiver 22 corresponds to a “GPS receiver”, and the posture estimation unit 24 includes “observation value calculation means”, “error estimation means”, and The storage device 26 corresponds to a “storage device”, and the SVT determination unit 30 and the wireless unit 40 correspond to a “theft determination notification unit”.

  Furthermore, the observation value calculation process shown in FIG. 8 corresponds to the process as the function of the “observation value calculation means”, and the combined process with the GPS shown in FIG. 7 corresponds to the process as the function of the “error estimation means”. The vehicle attitude calculation process shown in FIG. 6 corresponds to a process as a function of “vehicle attitude estimation means”.

As described above, the present invention is not limited to the above-described embodiment, and can be implemented in various forms without departing from the technical scope thereof.
In the above embodiment, two types of offset errors, that is, a fluctuation error and a stable error, are stored. However, three or more types of offset errors may be stored.

DESCRIPTION OF SYMBOLS 10 ... Theft notification apparatus 20 ... Vehicle posture estimation apparatus 21 ... GPS antenna 22 ... GPS receiver 23 ... Acceleration sensor 24 ... Posture estimation part 25 ... Real time clock (RTC)
26 ... Storage device 30 ... SVT determination unit 40 ... Wireless unit 50 ... Wireless communication service provider (MNO)
60 ... Service Provider (TSP)
70 ... GPS satellite

Claims (8)

An acceleration sensor for detecting acceleration generated in the vehicle;
A GPS receiver that outputs GPS data based on GPS positioning data;
An observation value calculating means for calculating an observation value from sensor data based on acceleration obtained by the acceleration sensor and the GPS data;
An error estimation unit that estimates an offset error by a Kalman filter using an observation value calculated by the observation value calculation unit, using an offset error of the acceleration sensor as a state quantity;
Vehicle attitude estimation means for estimating a vehicle attitude using the offset error estimated by the error estimation means;
A vehicle attitude estimation device comprising: The vehicle posture estimation apparatus according to claim 1,
The error estimation means calculates a mounting angle error of the acceleration sensor based on the gravitational acceleration when the vehicle is stopped and the acceleration in the vehicle traveling direction when the vehicle is traveling,
The vehicle attitude estimation device estimates the vehicle attitude using the attachment angle error and the offset error. The vehicle posture estimation apparatus according to claim 1 or 2,
The error estimation means stores the offset error in a storage device,
The vehicle attitude estimation device estimates the vehicle attitude using the offset error stored in the storage device even after power supply to the vehicle is stopped. In the vehicle posture estimation device according to claim 3,
The error estimation means stores a plurality of offset errors in the storage means based on an elapsed time since the start of power supply to the vehicle.
The vehicle posture estimation unit is configured to estimate the vehicle posture selectively using the plurality of offset errors based on an elapsed time after power supply to the vehicle is stopped. In the vehicle posture estimation device according to any one of claims 1 to 4,
The vehicle posture estimation device, wherein the error estimation means estimates the offset error when a vehicle speed based on the GPS positioning data exceeds a predetermined speed. In the vehicle posture estimation device according to any one of claims 3 to 5,
The error estimation means stores the offset error in the storage device when an error covariance value falls below a predetermined threshold value. The vehicle posture estimation device according to claim 6,
When the GPS data cannot be acquired, or when the vehicle speed based on the GPS data is equal to or lower than a predetermined speed, the error estimation unit determines an error covariance value when a certain time has elapsed since the previous estimation of the offset error. A vehicle attitude estimation device characterized by performing recalculation. The vehicle posture estimation device according to any one of claims 1 to 7,
A theft determination notification means for determining the presence or absence of a theft accident and notifying the outside based on the vehicle posture estimated by the vehicle posture estimation means of the vehicle posture estimation device;
A theft notification device comprising: