JP2019196945A - Failure detection system and program, and vehicle attitude estimation system and program - Google Patents

Abstract

To determine failed measuring means, of three pieces of measuring means, even when there are the three pieces of measuring means for estimating the attitude of a vehicle.SOLUTION: The failure detection system includes: three or more sensors 10 for measuring the same type of attitudes of a vehicle C; an attitude calculation unit 20 for calculating the attitude of the vehicle C on the basis of the measured values of the sensors 10; a difference calculation unit 30 for calculating the difference between the attitudes measured respectively by the three or more sensors 10; and a failure detection unit 50. On the basis of the difference, the failure detection unit determines, as a failed sensor 10', a sensor 10 that has calculated the attitude having a difference from the attitude calculated on the basis of the measured values by the other plurality of sensors 10, of the three or more sensors 10.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a failure detection system and program for detecting a failure of a measuring means, and a vehicle posture estimation system and program for estimating the posture of a vehicle using the measuring means.

  In order to perform automatic driving, driving assistance, and vehicle control in a vehicle, it is necessary to estimate the posture of the vehicle (eg, roll angle, pitch angle, yaw angle). For this purpose, the vehicle is provided with a sensor that measures a physical quantity according to the attitude of the vehicle and outputs the measured value. For example, a gyro sensor is used to measure the attitude of the vehicle, and the attitude of the vehicle is estimated based on the measurement value of the gyro sensor.

  When a failure occurs in such a sensor, it is impossible to estimate an accurate posture of the vehicle. Therefore, when a failure occurs in the sensor, it is necessary to detect it. Patent Document 1 discloses a method of detecting a failure of a gyro sensor by attaching four gyro sensors to the vehicle for estimating the attitude of the vehicle and comparing the angular velocities acquired by them. Specifically, in the method of Patent Document 1, by installing a gyro sensor on each of four slopes of a support base of a regular quadrangular pyramid (pyramid shape), a failure of any one of the gyro sensors is detected, and a correct angular velocity or Measure the angle.

JP 2005-221284 A

  However, according to the method described in Patent Document 1, it is not possible to measure the angular velocity in the direction when some of the gyro sensors fail with only three gyro sensors (see paragraph 0029 of Patent Document 1). ).

  Therefore, the present invention provides a fault detection system and program capable of identifying a faulty measurement unit, even when there are three measurement units for estimating the attitude of the vehicle, and three measurement units. It is an object of the present invention to provide a vehicle attitude estimation system and program capable of estimating the attitude of a vehicle even if any of them is out of order.

  In order to achieve the above object, the first aspect of the present invention is measured by three or more measuring means (10, 20) for measuring the same kind of posture of the vehicle, respectively, and the three or more measuring means. Difference calculation means (30) for calculating the difference between the postures, and based on the difference, a difference is calculated with respect to the postures measured by the plurality of other measurement means among the three or more measurement means. It is a fault detection system provided with the fault specification means (50) which specifies the said measurement means which measured the attitude | position which it has as a fault measurement means.

  In the second aspect of the present invention, three or more measuring means (10, 20) for measuring the same kind of attitude of the vehicle, respectively, and the difference between the attitudes respectively measured by the three or more measuring means are calculated. Difference calculating means (30) and attitude estimating means (60) for estimating the attitude of the vehicle based on the attitude measured by each of the three or more measuring means and the difference calculated by the difference calculating means. A vehicle attitude estimation system provided.

  According to a third aspect of the present invention, there is provided a failure for detecting the measurement means that has failed from the postures respectively measured by three or more measurement means that measure the same kind of posture of the vehicle by being executed in the arithmetic processing unit. A detection program, wherein the arithmetic processing unit is configured to calculate a difference between the postures respectively measured by the three or more measuring units, and based on the difference, the three or more It is a failure detection program that functions as a failure identification unit (50) that identifies the measurement unit that has measured a posture having a difference with respect to the posture measured by the plurality of other measurement units.

  According to a fourth aspect of the present invention, a vehicle attitude estimation is executed in the arithmetic processing unit to estimate the attitude of the vehicle from the attitudes respectively measured by three or more measuring units that measure the same kind of attitude of the vehicle. It is a program, and the arithmetic processing unit is measured by a difference calculating means (30) for calculating a difference in the posture measured by the three or more measuring means, and by the three or more measuring means, respectively. A vehicle posture estimation program that functions as posture estimation means (60) for estimating the posture of the vehicle based on the posture and the difference.

FIG. 1 is a block diagram showing a configuration of a vehicle attitude estimation system according to an embodiment of the present invention. FIG. 2 is a diagram illustrating a relationship between the vehicle according to the embodiment of the present invention, the field of view of the camera, and the vertical axis (when the road surface is horizontal). FIG. 3 is a diagram showing the relationship between the vehicle according to the embodiment of the present invention, the field of view of the camera, and the vertical axis (when the road surface is inclined from the horizontal). FIG. 4 is a diagram illustrating an example of a captured image when the road surface according to the embodiment of the present invention is inclined. FIG. 5 is a diagram showing a relationship between the vehicle according to the embodiment of the present invention, the field of view of the camera, and the vertical axis (when the road surface is inclined from the horizontal and the camera is displaced). FIG. 6 is a diagram illustrating an example of a captured image in the case of FIG. 5 according to the embodiment of the present invention. FIG. 7 is a diagram for explaining the difference in observation posture (when there is no axis deviation) according to the embodiment of the present invention. FIG. 8 is a diagram illustrating the difference in observation posture (when there is an axis deviation) according to the embodiment of this invention. FIG. 9 is a flowchart of vehicle posture estimation according to the embodiment of the present invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In addition, embodiment described below shows an example in the case of implementing this invention, Comprising: This invention is not limited to the specific structure demonstrated below. In carrying out the present invention, a specific configuration according to the embodiment may be adopted as appropriate.

  FIG. 1 is a block diagram showing a configuration of a vehicle attitude estimation system according to an embodiment of the present invention. The vehicle attitude estimation system 100 is configured in a vehicle and includes an IMU (Inertial Measurement Unit) 11, a gyro sensor 12, and a camera 13 (hereinafter collectively referred to as "sensor 10"). Each of these sensors 10 measures a physical quantity representing the attitude of the vehicle and outputs information representing the physical quantity.

  The IMU 11 is a sensor in which sensor groups for detecting angular velocity and acceleration are combined into one. Specifically, the IMU 11 includes three gyro sensors and three acceleration sensors, and the measured values are angular velocities around the three axes of the vehicle (that is, roll, pitch, yaw) and three axis directions (that is, the front-rear direction, the upper and lower directions). Direction, left and right). The IMU 11 may further include a geomagnetic sensor, an inclination sensor, and a GPS in order to correct a bias error of the gyro sensor or the acceleration sensor.

  The gyro sensor 12 is a single sensor provided separately from the IMU 11. The axis of the gyro sensor 12 faces the front-rear direction of the vehicle, and the gyro sensor 12 outputs the angular velocity of the roll angle as a measurement value.

  The camera 13 is fixed to the vehicle such that the optical axis faces the vehicle front-rear direction and the vertical axis coincides with the vehicle vertical direction. The camera 13 captures an image in front of the vehicle. The camera 13 outputs an image obtained by photographing (hereinafter referred to as “captured image”) as a measurement value.

  The vehicle posture estimation system 100 further includes a posture calculation unit 20, a difference calculation unit 30, a holding unit 40, a failure detection unit 50, a posture estimation unit 60, and an image correction unit 70. The posture calculation unit 20, the difference calculation unit 30, the failure detection unit 50, the posture estimation unit 60, and the image correction unit 70 are realized by an arithmetic processing device such as a processor executing a predetermined program, which will be described below. Perform the action. The holding unit 40 includes a storage device such as a memory accessible from the arithmetic processing device.

  Note that the posture calculation unit 20, the difference calculation unit 30, the failure detection unit 50, the posture estimation unit 60, and the image correction unit 70 may be realized by one processor or a plurality of processors. The holding unit 40 may also be composed of a plurality of storage devices.

  Further, in the vehicle posture estimation system 100, the failure detection system 101 is configured by the sensor 10, the posture calculation unit 20, the difference calculation unit 30, the holding unit 40, and the failure detection unit 50. As will be described later, the failure detection unit 50 outputs information for specifying the failure sensor 10 ′ and information on the axis deviation angle ε of the failure sensor 10 ′ as an output of the failure detection system 101, and the posture estimation unit 60 The posture of the vehicle is estimated using the information, and the image correction unit 70 corrects the image using the information, but the use of the output of the failure detection system 101 is not limited to these.

  The attitude calculation unit 20 calculates the attitude of the vehicle based on information output from each sensor 10. Specifically, the attitude calculation unit 20 acquires the vehicle pitch angle, roll angle, yaw angular velocity, and longitudinal, lateral, and vertical accelerations from the IMU 11, and based on these, the vehicle pitch angle The angle, roll angle, and yaw angle are calculated. For example, for the roll angle, the posture estimation unit 20 calculates the current roll angle of the vehicle by integrating the angular velocity of the roll angle acquired from the IMU 11.

  Further, the posture calculation unit 20 calculates the roll angle of the vehicle based on the angular velocity of the roll angle output from the gyro sensor 12. Specifically, the posture calculation unit 20 calculates the current roll angle of the vehicle by acquiring the angular velocity of the roll angle of the vehicle from the gyro sensor 12 and integrating the angular velocity. Further, the posture calculation unit 20 calculates the roll angle of the vehicle based on the image output from the camera 13.

  FIG. 2 is a diagram illustrating a relationship between the vehicle according to the embodiment of the present invention, the field of view of the camera, and the vertical axis (when the road surface is horizontal). As described above, the camera 13 is attached to the vehicle C such that the vertical direction (that is, the direction of the vertical axis 132) coincides with the vertical direction of the vehicle C. Further, the optical axis of the camera 13 faces the front-rear direction of the vehicle C.

  As shown in FIG. 2, the field of view 131 of the camera 13 is a rectangle, the upper side and the lower side are both parallel to the horizontal direction of the real space, and the left side and the right side are the vertical direction of the real space (that is, the direction of gravity). Parallel to R1. Therefore, when the vehicle C is traveling on a horizontal road surface R with no inclination, in the captured image, the left-right direction matches the horizontal direction of the real space, and the vertical direction matches the vertical direction of the real space. ing.

  FIG. 3 is a diagram showing the relationship between the vehicle according to the embodiment of the present invention, the field of view of the camera, and the vertical axis (when the road surface is inclined from the horizontal). As shown in FIG. 3, when the road surface R is inclined from the horizontal by an angle θ, the field of view 131 and the vertical axis 132 of the camera 13 are inclined by the angle θ with respect to the vertical axis R1. However, the structure S is built along the vertical axis R1 even if the road surface R is inclined. The attitude calculation unit 30 uses this to calculate the attitude of the vehicle (that is, the roll angle) from the captured image.

  FIG. 4 is a diagram illustrating an example of a captured image when the road surface according to the embodiment of the present invention is inclined. As shown in FIG. 4, when the road surface R is inclined by the angle θ, the portion in the structure portion S having a portion extending in the vertical direction appears to be inclined by the angle θ in the captured image. Examples of the building S having a portion extending in the vertical direction include a building and a pole.

  The posture calculation unit 20 has a function as an image processing unit in order to obtain the roll angle of the vehicle from the captured image. The image processing unit first generates an edge image indicating an edge of the image by performing image processing on the captured image. The image processing unit further extracts a straight edge having a predetermined length or more from the edge image and obtains the direction thereof.

  As a method for extracting a straight edge, for example, Hough transform (see, for example, Ballard, DH 1981. Generalizing the Hough transform to detect arbitrary patterns. Pattern Recognition, 13 (2): 111-122.), LSD (Line Segment Detector) algorithm (see, for example, von Gioi, RG, Jakubowicz, J., Morel, JM, Randall, G .: LSD: a Line Segment Detector. Image Processing On Line (2012)) can be used. Moreover, in order to improve detection capability, a method combined with edge detection may be used (for example, Canny, J., A Computational Approach To Edge Detection, IEEE Trans. Pattern Analysis and Machine Intelligence, 8 (6): 679. -698, 1986.). Furthermore, the accuracy can be further improved by detecting the image region of the building as preprocessing for extracting the linear edge. In this case, a method based on machine learning using a neural network or the like can be used for building detection.

  The posture calculation unit 20 calculates the angle formed by the extracted edge direction (that is, the vertical direction in the real space) with respect to the vertical direction of the captured image as the roll angle of the vehicle.

  FIG. 5 is a diagram showing a relationship between the vehicle according to the embodiment of the present invention, the field of view of the camera, and the vertical axis (when the road surface is inclined from the horizontal and the camera is displaced). In the case of FIG. 5, the road surface R is inclined at an angle θ, and the vertical axis 132 of the camera 13 is an angle ε about the axis in the front-rear direction of the vehicle C from the original direction (that is, around the optical axis of the camera 13). It's off.

  FIG. 6 is a diagram illustrating an example of a captured image in the case of FIG. 5 according to the embodiment of the present invention. As shown in FIG. 4, since the road surface R is originally inclined at the angle θ, the structure S should be inclined at the angle θ in the captured image. However, in the case of FIG. Since the camera 13 is inclined at an angle ε, the inclination of the structure S in the captured image is θ + ε.

  As described above, the attitude calculation unit 20 calculates the attitude (that is, the roll angle) of the vehicle based on the measurement value of the IMU 11, and calculates the attitude of the vehicle (that is, the roll angle) based on the measurement value of the gyro sensor 12. The vehicle attitude (that is, the roll angle) is calculated based on the measured value (that is, the captured image) of the camera 13.

  As described above, the IMU 11 and the attitude calculation unit 20 that calculates the attitude of the vehicle based on the information from the IMU 11 constitute one measuring unit that measures the attitude of the vehicle. The gyro sensor 12 and the gyro sensor The posture calculation unit 20 that calculates the posture of the vehicle based on the information from 12 constitutes another measuring means for measuring the posture of the vehicle. The camera 13 and the vehicle based on the information input from the camera 13 are configured. Still further measuring means for measuring the attitude of the vehicle is configured by the attitude calculation unit 20 that calculates the attitude of the vehicle.

  All of these three measuring means measure the angle of the vehicle around the same axis as the same kind of posture of the vehicle. Specifically, in the case of the present embodiment, each of the three measuring units measures the rotation angle around the axis extending in the front-rear direction of the vehicle C, that is, the roll angle, as the same kind of posture of the vehicle.

  In the present embodiment, one posture calculation unit 20 calculates the posture based on the measurement value of the sensor 10 for all the three sensors 10, but instead of this, for each sensor 10, the sensor 10 Means for calculating the attitude of the vehicle based on the measured values may be provided. Or each sensor 10 may be provided with the function of said attitude | position calculation part 20, and may output the attitude | position of a vehicle. In this case, the sensor 10 constitutes a measuring means.

  As described above, the measuring means including the sensor 10 and the attitude calculation unit 20 has the attitude based on information from the IMU 11 (that is, the roll angle) and the attitude based on information from the gyro sensor 12 (that is, as the same kind of attitude of the vehicle). , Roll angle) and a posture based on information from the camera 13 (that is, roll angle) are measured. Hereinafter, the posture calculated by the posture calculation unit 20 from the measurement value of the sensor 10 is referred to as “observation posture”. When the axis of the sensor 10 is deviated, the observation posture is calculated including the axis deviation angle.

Assuming that the observation posture calculated from the measurement value of the IMU 11 is θ1, the observation posture calculated from the measurement value of the gyro sensor 12 is θ2, and the observation posture calculated from the captured image of the camera 13 is θ3, these are true values θ In addition to the above, there is a possibility that the observation noise Δθ and the axis deviation angle ε of the sensor 10 may be included, which are generally expressed by the following equations (1) to (3).
θ1 = θ + Δθ1 + ε1 (1)
θ2 = θ + Δθ2 + ε2 (2)
θ3 = θ + Δθ3 + ε3 (3)

  The posture calculation unit 20 outputs the calculated observation postures θ1 to θ3 to the difference calculation unit 30 and the holding unit 40.

  The difference calculation unit 30 calculates the difference between the three observation postures obtained by the posture calculation unit 20. Specifically, the difference calculation unit 30 calculates the difference (θ1−θ2) between the observation posture θ1 based on the information from the IMU 11 and the observation posture θ2 based on the information from the gyro sensor 12, and information from the gyro sensor 12 The difference (θ2-θ3) between the observation posture θ2 based on the information from the camera 13 and the observation posture θ3 based on information from the camera 13 is calculated, and the observation posture θ3 based on the information from the camera 13 and the observation posture θ1 based on the information from the IMU 11 are calculated. The difference (θ3-θ1) is calculated.

  The difference calculation unit 30 outputs the calculated three differences to the holding unit 40 together with the time information. The holding unit 40 stores the three differences and the time information in association with each other.

  The failure detection unit 50 reads the difference stored in the holding unit 40 and identifies the sensor 10 in which the failure has occurred as the failure sensor 10 ′. Specifically, the failure detection unit 50 identifies one sensor 10 having a calculated posture different from the posture calculated from the other two sensors 10 as the failure sensor 10 ′.

  The failure detection unit 50 calculates not only the difference between observation postures (that is, instantaneous values) obtained at one time but also an average value (that is, time average) of differences between a plurality of observation postures obtained over a certain time width. Thus, the difference between each of the three observation postures is calculated. That is, the difference calculation unit 30 statistically processes the difference within a predetermined time, thereby reducing the influence of the observation noise and obtaining the difference due to the influence of the axis deviation. The failure detection unit 50 includes the latest difference, and calculates the average of a plurality of differences from the latest difference up to a point in the past by a predetermined time as a time average.

  If any sensor 10 has no axis deviation and the axial tone is perfect, the axis deviation angles ε1, ε2, and ε3 in equations (1) to (3) are all zero. In addition, since the observation noises Δθ1, Δθ2, and Δθ3 in the expressions (1) to (3) are independent of time, these time averages can be regarded as zero.

  Now, when there is no axis deviation (that is, when ε1 = 0, ε2 = 0, ε3 = 0), the difference (θ1-θ2) (= θ + Δθ1 + ε1- (θ + Δθ2 + ε2)) is Δθ1-Δθ2 when taking a time average. Since this component becomes 0, θ−θ = 0. Similarly, for the difference (θ2−θ3) and the difference (θ3−θ1), when there is no axis shift, the time average is 0.

  FIG. 7 is a diagram for explaining the difference in observation posture (when there is no axis deviation) according to the embodiment of the present invention. 7A shows the difference (θ1−θ2) in the observation posture between the IMU 11 and the gyro sensor 12, and FIG. 7B shows the observation posture between the gyro sensor 12 and the camera 13. The difference (θ2−θ3) is shown, and FIG. 7C shows the difference in the observation posture (θ3−θ1) between the camera 13 and the IMU 11. In the example of FIG. 7, a case is shown in which none of the sensors 10 is misaligned (that is, ε1 = 0, ε2 = 0, ε3 = 0).

  As shown in FIG. 7, when the difference at one time (that is, the instantaneous value) is viewed, the difference is not necessarily 0 due to observation noise that occurs randomly in each sensor 10, but when taking the time average, the axis When there is no deviation, the observation noises generated at random are canceled out, and the difference becomes 0 as an average value.

  On the other hand, when an axis deviation has occurred, taking the time average of each difference, the observation noises cancel each other, and the influence ε of the axis deviation remains. For example, if an axis deviation ε1 occurs in the IMU 11, the time average of the difference (θ1-θ2) is ε1, the time average of the difference (θ2-θ3) is 0, and the time average of the difference (θ3-θ1) is -Ε1. Thus, since the time average of the difference (θ1−θ2) is ε1 and the time average of the difference (θ3−θ1) is −ε1, it can be seen that a shift of ε1 occurs in θ1.

  FIG. 8 is a diagram illustrating the difference in observation posture (when there is an axis deviation) according to the embodiment of this invention. Similar to the example in FIG. 7, FIG. 8A shows the difference (θ1−θ2) in the observation posture between the IMU 11 and the gyro sensor 12, and FIG. 8B shows the gyro sensor 12 and the camera. FIG. 8C shows the observation difference (θ3-θ1) between the camera 13 and the IMU 11. In the example of FIG. 8, a case where an axis deviation occurs in the gyro sensor 12 is shown.

  When an axis deviation occurs in the gyro sensor 12, the difference in observation posture between the IMU 11 and the gyro sensor 12 shown in FIG. 8A does not become zero, and the gyro sensor 12 and the camera 13 shown in FIG. The difference in the observation posture between and does not become zero. From this, it can be seen that an axis deviation occurs in the gyro sensor 12.

  The failure detection unit 50 outputs information on the specified failure sensor 10 ′ and its axis deviation angle ε to the posture estimation unit 60 and the image correction unit 70. The posture estimation unit 60 determines the vehicle posture based on the observation postures (that is, roll angles) θ1, θ2, and θ3 calculated by the posture calculation unit 20 and information on the failure sensor 10 ′ and its axis deviation angle ε. The true value θ is estimated. At this time, the posture estimation unit 60 reads out a plurality of observation postures θ1, θ2, and θ3 having a predetermined time width held in the holding unit 40, takes the respective time averages, and observes the observation postures θ1, Let θ2 and θ3.

  The posture estimation unit 60 estimates that one of the observation postures θ1, θ2, and θ3, or an average thereof is a true value θ when any sensor 10 has no axis deviation. If any sensor 10 has an axis deviation ε, the attitude estimation unit 60 corrects the observation attitude of the failure sensor 10 ′ so as to cancel the axis deviation ε, and uses the corrected observation attitude. To estimate the true value θ. Alternatively, when any sensor 10 has an axis deviation, the attitude estimation unit 60 ignores the observation attitude of the failure sensor 10 ′ and uses only the observation attitude of the sensor 10 in which no axis deviation occurs. The true value θ of the posture may be estimated.

  When receiving information from the failure detection unit 50 indicating that the camera 13 has been identified as the failure sensor 10 ′, the image correction unit 70 corrects the captured image based on the axis deviation angle ε 3 of the camera 13. The axis deviation angle ε3 is obtained from the attitude difference (θ2−θ3) (= −ε3) and / or the attitude difference (θ3−θ1) (= ε3) in the failure detection unit 50.

  FIG. 9 is a flowchart of vehicle posture estimation according to the embodiment of the present invention. First, each sensor 10 performs sensing and acquires a measured value (step S91). The attitude calculation unit 20 calculates the observation attitude of the vehicle based on the measured values of the sensors 10 and stores the calculated observation attitude in the holding unit 40 (step S92). The difference calculation unit 30 calculates the difference between the observation postures at the same time and stores it in the holding unit 40 (step S93).

  The failure detection unit 50 detects the failure sensor 10 ′ based on each difference, and obtains the axis deviation angle ε (step S94). If there is no failure sensor 10 '(NO in step S95), failure detection unit 50 estimates the true value of the posture by taking the time average of each observation posture held in holding unit 40 (step S96). . If there is a failure sensor 10 ′ (YES in step S95), the failure detection unit 50 corrects the observation posture of the failure sensor 10 ′ (step S97) and estimates the true value of the posture (step S96). ).

  As described above, according to the failure detection system of the embodiment of the present invention, a sensor failure can be detected if at least three measuring means for measuring the same kind of posture of the vehicle are provided. According to the vehicle posture estimation system of this form, if at least three measuring means for measuring the same kind of posture of the vehicle are provided, the posture of the vehicle can be correctly estimated even if any of the sensors has failed.

  In the above embodiment, the attitude calculation unit 20 calculates the roll angle of the vehicle as the same kind of attitude of the vehicle based on the measured values of the IMU 11, the gyro sensor 12, and the camera 13. When the vehicle is facing the left-right direction of the vehicle, the pitch angle of the vehicle can be measured based on the captured image. In this case, by providing two or more sensors 10 that can calculate the pitch angle, it is possible to configure three or more measuring means that can measure the pitch angle of the vehicle as the same kind of posture of the vehicle.

  Further, in the same kind of posture measured by three or more measuring means, a fixed (that is, known) bias or offset may exist for some measuring means. In this case, the posture calculation unit 20 and / or the difference calculation unit 30 can consider a bias or an offset so that the time average of the difference can be zero when there is no axis deviation, and the present invention is applied. Is possible.

  In the above-described embodiment, one of the sensors 10 is the camera 13, but the present invention is not limited to this, and three or more gyro sensors may constitute three or more measuring means. Three or more measuring means may be configured by two or more cameras.

  In the above embodiment, the time average of a plurality of differences obtained by calculating the difference at each time is taken. Instead, the difference calculation unit 30 calculates the time average of the posture for each sensor 10. You may calculate those differences after taking. Further, instead of taking the time average of differences or postures, statistical processing may be performed using accumulated values of differences or postures.

11 IMU, 12 gyro sensor, 13 camera, 20 attitude calculation unit,
30 difference calculation unit, 40 holding unit, 50 failure detection unit, 60 posture estimation unit,
70 Image correction unit, 131 field of view, 132 vertical axis

Claims (13)

Three or more measuring means (10, 20) for measuring the same kind of posture of the vehicle,
Difference calculating means (30) for calculating a difference between the postures respectively measured by the three or more measuring means;
A failure that identifies, as a failure measurement unit, the measurement unit that measures the posture having a difference with respect to the posture measured by the plurality of other measurement units among the three or more measurement units based on the difference. Identifying means (50);
A failure detection system.   The failure detection system according to claim 1, wherein the failure identification unit identifies the failure measurement unit based on the accumulated difference.   The failure detection system according to claim 1, further comprising a correction unit that corrects the posture measured by the failure measurement unit according to the difference.   2. The at least one of the three or more measuring units includes a camera fixed to the vehicle and an image processing unit that measures the posture based on an image obtained by photographing with the camera. The failure detection system in any one of -3.   The image processing means detects a structure along the vertical direction included in the image, and calculates a roll angle of the vehicle as the posture based on the orientation of the structure in the image. The described failure detection system.   The image correction unit (70) according to claim 4 or 5, further comprising an image correction unit (70) for correcting an image obtained by photographing of the camera based on the difference when the camera is specified as the failure measurement unit. Failure detection system.   The failure detection system according to claim 1, wherein the three or more measuring units measure an angular velocity around a coaxial axis as the same kind of posture of the vehicle.   The failure detection system according to claim 1, wherein the three or more measuring units measure any one of a pitch angle, a yaw angle, and a roll angle of the vehicle as the same kind of posture of the vehicle. Three or more measuring means (10, 20) for measuring the same kind of posture of the vehicle,
Difference calculating means (30) for calculating a difference between the postures respectively measured by the three or more measuring means;
Attitude estimation means (60) for estimating the attitude of the vehicle based on the attitude measured by the three or more measurement means and the difference calculated by the difference calculation means;
A vehicle posture estimation system. A failure that identifies, as a failure measurement unit, the measurement unit that measures the posture having a difference with respect to the posture measured by the plurality of other measurement units among the three or more measurement units based on the difference. Further comprising a specifying means (50),
The posture estimating means corrects the posture measured by the failure measuring means;
The vehicle posture estimation system according to claim 9. A failure that identifies, as a failure measurement unit, the measurement unit that measures the posture having a difference with respect to the posture measured by the plurality of other measurement units among the three or more measurement units based on the difference. Further comprising a specifying means (50),
The attitude estimation means estimates the attitude of the vehicle ignoring the attitude measured by the failure measurement means;
The vehicle posture estimation system according to claim 9. A failure detection program for detecting the measurement means that has failed from the postures respectively measured by three or more measurement means that measure the same kind of posture of the vehicle by being executed in the arithmetic processing device, Equipment
Difference calculating means (30) for calculating the difference between the postures respectively measured by the three or more measuring means, and a plurality of other measurements of the three or more measuring means based on the difference A fault identification unit (50) for identifying the measurement unit that has measured a posture having a difference from the posture measured by the unit;
Failure detection program that functions as A vehicle attitude estimation program for estimating the attitude of a vehicle from the attitudes respectively measured by three or more measuring means for measuring the same kind of attitude of the vehicle by being executed in the arithmetic processing apparatus, wherein the arithmetic processing apparatus The
A difference calculating means (30) for calculating a difference between the postures measured by the three or more measuring means; and a posture of the vehicle based on the posture and the difference respectively measured by the three or more measuring means. Posture estimation means (60) for estimating the posture;
As a vehicle posture estimation program.

Images