Position estimation device

Abstract

Position estimation device (1) with: - an image capture unit (11) configured to capture a plurality of images from a plurality of cameras (20a to 20d) mounted on a vehicle (2) to capture images around the vehicle (2), wherein the plurality of cameras (20a to 20d) are mounted to form a plurality of overlapping areas (3, 4, 5, 6), the overlapping areas (3, 4, 5, 6) being areas in which image areas partially overlap each other; - a detection unit (11, 12, S100 to S110) configured to detect at least one feature portion, which is a portion indicating a predetermined feature in an area on each of a plurality of images each including appearances of the overlapping areas (3, 4, 5, 6), the plurality of images being detected by the image detection unit (11) at approximately the same time; - an estimation unit (12, 13, S112 to S118) configured to calculate, for each of the overlapping areas (3, 4, 5, 6), an amount of positional offset between the feature portions acquired from the plurality of images each including appearances of the overlapping areas (3, 4, 5, 6), and to estimate a predetermined physical quantity associated with an attitude of the vehicle (2) based on a difference in the amount of offset calculated for each of the plurality of overlapping areas (3, 4, 5, 6); - a position information output unit (14, S130) configured to output information based on a physical quantity estimated by the estimation unit (12, 13, S112 to S118) as position information, which is information indicating a position of the vehicle (2); - a reliability determination unit (14, S120, S122) configured to determine a reliability of a gyro sensor (21) provided in the vehicle (2) based on a result of a comparison of a posture of the vehicle (2) indicated by a physical quantity estimated by the estimation unit (12, 13, S112 to S118) with a measurement associated with a posture of the vehicle (2) derived from the gyro sensor (21); and - a reliability information output unit (14, S124) configured to output unreliability information, which is information indicating that the reliability of the gyro sensor (21) is determined to be unreliable when the reliability of the gyro sensor (21) is determined to be unreliable by the reliability determination unit (14, S120, S122).

Description

[Technical field]

[0001] The present invention relates to an attitude estimating apparatus that estimates an attitude of a vehicle based on images taken by cameras mounted on the vehicle. [State of the art]

[0002] JP 2012-026 992 A describes a technique for estimating a pitch angle, which is one of rotation angles associated with an attitude of the vehicle, based on multiple images taken by a vehicle camera at different times.

[0003] The technique according to JP 2012-026992 A allows estimation of the vehicle's pitch angle based on images captured by the vehicle's camera, but does not consider estimating other rotation angles, such as roll angle or yaw angle. As a result of detailed research, the inventors of the present disclosure have found a problem inherent in the technique according to JP 2012-026992 A. Specifically, the technique according to JP 2012-026992 A is insufficient to estimate a comprehensive attitude including the vehicle's pitch angle, roll angle, yaw angle, and the like.

[0004] US 2015 / 0 332 098 A1 further discloses a system and method designed to estimate the dynamics of a mobile platform by matching feature points in overlapping images from cameras on the platform, such as cameras in a surround-view camera system in a vehicle. The method comprises the steps of identifying overlap image regions for any two cameras in the surround-view camera system, identifying common feature points in the overlap image regions, and determining that the common feature points are not located at the same location in the overlap image regions. The method further comprises the steps of estimating three-degree-of-freedom vehicle dynamics parameters from the matching between the common feature points, and estimating the vehicle dynamics of one or more of the pitch, roll, and height variations using the vehicle dynamics parameters.

[0005] JP 2007-278 871 A discloses an apparatus for calculating a movement amount of a moving object based on processing images taken by cameras provided for the moving object.

[0006] It is an object of the present disclosure to provide a technique with which the reliability of information associated with a position of the vehicle can be increased.

[0007] The problem is solved by the subject matter of the independent claims.

[0008] According to the invention, in a case where multiple cameras are mounted on a vehicle, a change in the attitude of the vehicle results in a displacement of the cameras. This displacement of the cameras also results in a positional offset between the objects appearing in the overlapping image areas of the vehicle, depending on the direction or degree of the change in the vehicle's attitude. In this regard, the attitude estimation device of the present disclosure allows multiple cameras to capture images with the image areas partially overlapping and detects feature portions included in the respective overlapping areas of the images. Subsequently, based on the amount of positional offset between the feature portions for each of the overlapping areas, an attitude of the vehicle is estimated. In this way, a technique that can estimate various physical quantities associated with an attitude of the vehicle can be realized. [Brief description of the drawings]

[0009] The objects, features, and advantages of the present disclosure will become more apparent from the following description with reference to the accompanying drawings. The drawings are briefly described below. Fig. 1 is a block diagram illustrating a configuration of a rotation angle estimation device according to an embodiment. Fig. Figure 2 shows a diagram illustrating mounting positions of cameras and image areas of the cameras. Fig. 3 shows a flowchart illustrating a process executed by a rotation angle estimating device. Fig. 4 shows a flowchart illustrating the process performed by the rotation angle estimation device. Fig. 5 shows a set of figures illustrating a change in captured images due to a change in the attitude of the vehicle. [Description of the embodiments]

[0010] Hereinafter, an embodiment of the present disclosure will be described with reference to the drawings. It should be noted that the present disclosure is not limited to the following embodiment, but can be implemented in various modes. (Configuration of rotation angle estimator)

[0011] The present embodiment provides a rotation angle estimation device 1, which is an electronic control unit installed in a vehicle 2. The rotation angle estimation device 1 corresponds to the attitude estimation device of the present disclosure. As shown in Fig. 1, the rotation angle estimation device 1 is connected to cameras 20a, 20b, 20c, 20d, a gyro sensor 21 and a temperature sensor 22 installed in the vehicle 2.

[0012] As in Fig. 2, the four cameras 20a to 20d, which serve as the front camera, rear camera, left camera, and right camera, are mounted at the front, rear, left, and right sides of the vehicle 2 so that the surroundings of the vehicle can be imaged. The cameras 20a to 20d are used to capture images around the vehicle and, as indicated by the dashed lines in Fig. 2, corresponding image areas in the front, rear, right, and left directions of the vehicle 2. The image areas of the cameras 20a to 20d include corresponding overlap areas 3, 4, 5, and 6 at the front left, front right, rear left, and rear right of the vehicle 2. In these overlap areas, adjacent image areas of the cameras 20a to 20d overlap.

[0013] The mounting positions and settings of the cameras 20a to 20d are preset with respect to the vehicle 2, so that the image areas of the cameras 20a to 20d are formed as described above. It is assumed that the actual mounting positions of the cameras 20a to 20d in the manufacturing plants, maintenance workshops, or the like are adjusted based on the preset mounting positions and settings (e.g., image directions).

[0014] Below is again Fig. 1. The gyro sensor 21 is a known measuring device that detects a rotational angular velocity (e.g., yaw rate, pitch rate, and / or roll rate) of the vehicle 2. The temperature sensor 22 is a known measuring device that measures an ambient temperature of the vehicle 2. The measurement results of the gyro sensor 21 and the temperature sensor 22 are input to the rotation angle estimation device 1.

[0015] The rotation angle estimation device 1 is an information processing unit mainly composed of a central processing unit (CPU), a random access memory (RAM), a read-only memory (ROM), a semiconductor memory such as a flash memory, an input / output interface, and the like (all not shown). The rotation angle estimation device 1 is embodied, for example, by a microcontroller or the like that comprehensively functions as a computer system. The functions of the rotation angle estimation device 1 are implemented by the CPU, which executes the programs stored on a non-volatile tangible recording medium such as a ROM or a semiconductor memory. The rotation angle estimation device 1 may be configured by one or more microcontrollers.

[0016] The rotation angle estimation device 1 has a function of estimating a physical quantity, such as a rotation angle or a rotation angular velocity, indicative of a posture of the vehicle based on the images captured by the cameras 20a to 20d. As components for the function, the rotation angle estimation device 1 includes a feature point detection unit 11, a posture estimation unit 12, a rotation angular velocity calculation unit 13, and a reliability determination unit 14. These components constituting the rotation angle estimation device 1 can be implemented not only in software but also in hardware for part or all of the components by combining a logic circuit, an analog circuit, and the like.

[0017] The rotation angle estimation device 1 has a function of generating bird's-eye view images of the vehicle 2 by converting the plurality of images captured by the cameras 20a to 20d into view angles using the mounting positions and settings of the cameras 20a to 20d as camera parameters. The camera parameters are obtained, for example, by digitizing the mounting positions of the cameras 20a to 20d on the vehicle 2 and the mounting angles of the cameras in the three-axis direction, that is, the longitudinal, lateral, and vertical directions of the vehicle 2. When converting the images captured by the cameras 20a to 20d into view angles, the rotation angle estimation device 1 uses conversion data determined based on the camera parameters. (Process performed by the rotation angle estimator)

[0018] The following describes a process executed by the rotation angle estimation device 1 with reference to the Fig. 3 and Fig. 4. This process is repeated in a predetermined control cycle.

[0019] First of all, the Fig. 3. In step S100, the rotation angle estimation device 1 acquires a plurality of images taken by the cameras 20a to 20d at approximately the same time. Subsequently, the rotation angle estimation device 1 acquires a position of at least one feature point from each of the bird's-eye view images obtained by converting the plurality of images into view angles. The feature point refers to a portion of the image in which a specific object appears on the road surface. The object to be acquired as the feature point may be one that can be easily distinguished from the road surface in each image. Examples of such an object include a dividing line or road marking with paint, posts or stones, and a manhole cover.

[0020] In step S102, the rotation angle estimator 1 corrects a positional offset between the feature points acquired in step S100. The positional offset between the feature points is attributed to the offset between the camera parameters, the actual mounting positions and settings of the cameras 20a and 20b, and the center of gravity imbalance of the vehicle 2. Known values ​​are applied to the mounting positions and settings of the cameras 20a and 20b, as well as the center of gravity imbalance. The known values ​​are obtained, for example, by measurements performed separately using a known method based on the images captured by the cameras 20a and 20b. Steps S100 and S102 correspond to processing as the feature point acquisition unit 11.

[0021] In step S104, the rotation angle estimator 1 extracts at least one feature point included in an image area corresponding to the front-left overlap area 3 from one or more feature points acquired in step S100 for each of the bird's-eye view images of the front camera 20a and the left camera 20c. In step S106, the rotation angle estimator 1 extracts a feature point included in an image area corresponding to the front-right overlap area 4 from the feature points acquired in step S100 for each of the bird's-eye view images of the front camera 20a and the right camera 20d.

[0022] In step S108, the rotation angle estimator 1 extracts a feature point included in an image area corresponding to the rear-left overlap area 5 from the feature points acquired in step S100 for each of the bird's-eye view images of the rear camera 20b and the left camera 20c. In step S110, the rotation angle estimator 1 extracts a feature point included in an image area corresponding to the rear-right overlap area 6 from the feature points acquired in step S100 for each of the bird's-eye view images of the rear camera 20b and the right camera 20d.

[0023] In step S112, the turning angle estimator 1 estimates a turning angle (ie, vehicle turning angle) indicating an attitude of the vehicle 2 based on the positions of the feature points extracted in steps S104 to S110 in the bird's-eye view images corresponding to the overlapping areas 3 to 6, respectively. In the present embodiment, for example, a pitch angle, a roll angle, a yaw angle, or the like is understood as the vehicle turning angle estimated by the turning angle estimator 1.

[0024] When acceleration, deceleration, and / or rotational force are applied to the vehicle by rotation, the vehicle's attitude may change. The change in the vehicle's attitude results in a change in the positions of cameras 20a and 20b relative to the road surface, which is an object to be imaged. Therefore, a common feature point between the two cameras capturing an image with the same overlap region experiences a positional offset. The degree of this offset depends on the position of the overlap region relative to the vehicle, as well as the direction and degree of the attitude change.

[0025] Below are concrete examples with reference to Fig. 5 described. In Fig. 5 shows a bird's-eye view images taken by the left and right cameras 20c and 20d before a change in attitude occurs in the vehicle 2. b shows bird's-eye view images taken by the front and rear cameras 20a and 20b at approximately the same time as in a. c shows bird's-eye view images taken by the left and right cameras 20c and 20d when the pitch angle has changed so that the vehicle may lean forward due to rapid deceleration. d shows bird's-eye view images taken by the front and rear cameras 20a and 20b at approximately the same time as in c. In the examples of Fig. 5, dividing lines of the road that run along the direction of travel are recorded as feature points.

[0026] If the pitch angle has changed so that the vehicle can lean forward due to rapid deceleration, the front camera 20a is shifted more than the other cameras. In this case, as shown in Fig. 5, a comparatively large difference in the appearance of the dividing lines, as feature points, in the overlap areas 3 and 4 between the bird's eye view images of c taken by the left and right cameras 20c and 20d and the bird's eye view images of d taken by the front camera 20a.

[0027] If the attitude of the vehicle has changed so that it leans forward, the displacement of the rear camera 20b is small compared to the front camera 20a. Consequently, as shown in Fig. 5, there is not such a large difference in the appearance of the dividing lines, as feature points, in the overlapping areas 5 and 6 between the bird's eye view images of c taken by the left and right cameras 20c and 20d and the bird's eye view images of d taken by the rear camera 20b.

[0028] In this context, the turning angle estimator 1 compares the amount of offset between feature points in the front overlap areas 3 and 4 with the amount of offset between feature points in the rear overlap areas 5 and 6, respectively, and estimates a pitch angle based on the magnitude ratio in the amount of offset. Similarly, the turning angle estimator 1 can compare the amount of offset between feature points in the left overlap areas 3 and 5 with the amount of offset between feature points in the right overlap areas 4 and 6, respectively, and estimate a roll angle based on the magnitude ratio in the amount of offset. The turning angle estimator 1 estimates a yaw angle from the angle of the dividing line detected by each of the cameras 20a to 20d with respect to the longitudinal direction of the vehicle.

[0029] With reference to the Fig. 3, the processing in steps S104 to S112 corresponds to the processing as the attitude estimation unit 12. In the subsequent step S114, the rotation angle estimation device 1 determines whether a necessary amount of time series information on the vehicle rotation angle for a past predetermined period from the current point has been collected. If the required amount of time series information has not been collected (ie, NO in step S114), the rotation angle estimation device 1 allows the processing to proceed to step S118.

[0030] In step S118, the rotation angle estimator 1 accumulates the vehicle rotation angle obtained in step S112 in memory as time-series information. If it is determined that the required amount of time-series information has been collected (ie, YES in step S114), the rotation angle estimator 1 allows the processing to proceed to step S116.

[0031] In step S116, the rotation angle estimator 1 calculates a vehicle rotation angular velocity indicating a temporal change in the vehicle rotation angle. The vehicle rotation angular velocity is calculated based on the vehicle rotation angle obtained in step S112 and the time series information of the vehicle rotation angle collected for the elapsed predetermined period from the current point. The vehicle rotation angular velocity is calculated, for example, as a value obtained by dividing a time-series displayed change in the vehicle rotation angle by an elapsed time in the time series. In the present embodiment, for example, a pitch velocity, a roll velocity, a yaw rate, or the like are understood as the vehicle rotation angular velocity estimated by the rotation angle estimator 1.

[0032] The required amount of information about the vehicle rotation angle used to calculate a vehicle rotation angular velocity may be accumulated for at least one calculation, but a larger amount of information may also be accumulated. For example, if the amount of required information about the vehicle rotation angle is increased, the robustness in the result of calculating a vehicle rotation angular velocity to unexpected noise is correspondingly improved. However, this increase may impair the adaptability of the calculated vehicle rotation angular velocity to an abrupt change in the vehicle rotation angle. If the amount of required information about the vehicle rotation angle is decreased, this reduction may improve the adaptability of the calculated vehicle rotation angular velocity to an abrupt change in the vehicle rotation angle.However, this decrease may compromise the robustness of the vehicle angular velocity calculation result to unexpected noise. Thus, the robustness and responsiveness of the vehicle angular velocity calculation results are a trade-off, depending on the amount of required vehicle angular velocity information used to calculate the vehicle angular velocity.

[0033] After step S116, the rotation angle estimation device 1 executes the processing of step S118. It should be noted that the processing of steps S114 to S118 corresponds to the processing as the rotation angle calculation unit 13.

[0034] Below is the Fig.4. In step S120, the rotation angle estimator 1 calculates a difference between the vehicle angular velocity estimated in step S116 and the last measurement obtained from the gyro sensor 21. Subsequently, the rotation angle estimator 1 accumulates the calculated difference as a comparison result. In step S122, the rotation angle estimator 1 determines whether the difference between the estimated vehicle angular velocity and the measurement obtained from the gyro sensor 21 continues to be not less than a predetermined value, for the comparison results accumulated in step S120 for a lapsed predetermined period from the current point.

[0035] In step S122, the reliability of the gyro sensor 21 is determined. Specifically, if there is no significant difference between the vehicle rotational angular velocity estimated from the images of the cameras 20a to 20d and the vehicle rotational angular velocity measured by the gyro sensor 21, the gyro sensor 21 is determined to be functioning normally. If there is a certain difference, lasting for a predetermined period, between the vehicle rotational angular velocity estimated from the images captured by the cameras 20a to 20d and the vehicle rotational angular velocity measured by the gyro sensor 21, the reliability of the gyro sensor 21 is determined to be unreliable.

[0036] If the comparison result difference not smaller than a predetermined value persists for a predetermined period of time (i.e., YES in step S122), the rotation angle estimator 1 allows the processing to proceed to step S124. In step S124, the rotation angle estimator 1 outputs an error message or information indicating a failure of the gyro sensor 21 to an output device or the like (not shown) installed in the vehicle 2. In step S126, the rotation angle estimator 1 outputs the vehicle rotation angle estimated in step S112 and the vehicle rotation angular velocity calculated in step S116 as attitude information, i.e., information indicating an attitude of the vehicle 2.

[0037] If the difference in the comparison result, which is not less than a predetermined value, does not persist for a predetermined period of time (ie, NO in step S122), the rotation angle estimator 1 allows the processing to proceed to step S128. In step S128, the rotation angle estimator 1 determines whether the ambient temperature of the vehicle 2 is within an appropriate range based on the measurement detected by the temperature sensor 22.

[0038] The appropriate ambient temperature range may be, for example, a temperature range in which the performance of the gyro sensor 21 is ensured. It is generally known that the measurement performance of gyro sensors installed in vehicles or the like is degraded under extremely high or low temperature conditions. In this regard, in the configuration of the present embodiment, in step S122, it can be reliably determined whether to use the measurement of the gyro sensor 21 in accordance with the ambient temperature.

[0039] If the ambient temperature is outside the appropriate range (ie, NO in step S128), the rotation angle estimator 1 allows the processing to proceed to step S126. In step S126, the rotation angle estimator 1 outputs the vehicle rotation angle estimated in step S112 and the vehicle rotation angular velocity calculated in step S116 as attitude information, ie, information indicating an attitude of the vehicle 2.

[0040] If the ambient temperature is within the appropriate range (ie, YES in step S128), the rotation angle estimator 1 allows the processing to proceed to step S130. In step S130, the rotation angle estimator 1 outputs the vehicle rotation angular velocity detected by the gyro sensor 21 as attitude information, ie, information indicative of an attitude of the vehicle 2. (Beneficial effects)

[0041] According to the rotation angle estimating device 1 of the embodiment, the following advantageous effects are obtained.

[0042] The rotation angle estimator 1 detects feature points included in areas where overlapping regions occur from the bird's-eye view images captured by the plurality of cameras 20a to 20d whose image areas partially overlap. As a result of the detection, the rotation angle estimator 1 can estimate a vehicle rotation angle indicative of a vehicle's attitude based on the amount of offset between the feature points for each of the plurality of overlapping regions. Thus, the rotation angle estimator 1 can be an alternative to the gyro sensor 2. Otherwise, if the rotation angle estimator 1 is combined with the gyro sensor 21, the means for detecting a vehicle's attitude can be made redundant.

[0043] Furthermore, by comparing the vehicle attitude estimated based on the bird's-eye view images captured by the cameras 20a to 20d with the vehicle attitude measured by the gyro sensor 21, the reliability of the gyro sensor 21 can be determined. With this configuration, for example, an error message for the gyro sensor 21 can be output under conditions where the reliability of the gyro sensor 21 is determined to be unreliable. Furthermore, from a fail-safe perspective, the configuration enables control to disable the measurement of the gyro sensor 21.

[0044] Furthermore, the attitude information based on the images or the attitude information based on the measurement of the gyro sensor 21 can be selectively output according to the ambient temperature. With this configuration, a favorable portion of attitude information can be output according to the vehicle's surroundings. Thus, this configuration contributes to improving the measurement accuracy of the attitude information. (Correlation of components in the embodiment)

[0045] The feature point detection unit 11 corresponds to an example of the image acquisition unit and the detection unit. The attitude estimation unit 12 and the rotation angular velocity calculation unit 13 correspond to an example of the estimation unit. The reliability determination unit 14 corresponds to an example of the attitude information output unit, the reliability information determination unit, the reliability information output unit, and the environment determination unit. (Modifications)

[0046] The function of one component in the above-described embodiments may be distributed among multiple components, or the functions of multiple components may be performed by one component. Moreover, part of the configuration of each embodiment described above may be omitted. Furthermore, at least part of the configuration of one embodiment described above may be added to or replaced by the configuration of another embodiment described above. It should be noted that all modes included in the technical idea specified by the language of the claims should be embodiments of the present disclosure.

[0047] The above embodiment is described as an example of estimating a vehicle posture based on the images captured by the four cameras 20a to 20d. However, the present disclosure should not be limited to this example. A configuration may be such that a vehicle posture is estimated based on the images captured by more than or fewer than four cameras, as long as a plurality of overlapping areas are formed in which the image areas each partially overlap.

[0048] The present disclosure can be implemented in various modes, such as a system including the rotation angle estimating device 1 as a component, a program that enables a computer to function as the rotation angle estimating device 1, a non-transitory tangible recording medium on which this program is recorded, and a method for estimating a vehicle rotation angle.

Claims

An attitude estimation device (1) comprising:- an image acquisition unit (11) configured to acquire multiple images from multiple cameras (20a to 20d) mounted on a vehicle (2) to capture images around the vehicle (2), wherein the multiple cameras (20a to 20d) are mounted to form multiple overlapping regions (3, 4, 5, 6), wherein the overlapping regions (3, 4, 5, 6) are regions in which image regions partially overlap;- a detection unit (11, 12, S100 to S110) configured to detect at least one feature portion, which is a portion indicating a predetermined feature in an area, on each of multiple images each including appearances of the overlapping regions (3, 4, 5, 6), wherein the multiple images are acquired by the image acquisition unit (11) at approximately the same time;- an estimation unit (12, 13, S112 to S118) configured to calculate, for each of the overlapping areas (3, 4, 5, 6), an amount of positional offset between the feature portions acquired from the plurality of images each including appearances of the overlapping areas (3, 4, 5, 6), and to estimate a predetermined physical quantity associated with a posture of the vehicle (2) based on a difference in the amount of offset calculated for each of the plurality of overlapping areas (3, 4, 5, 6); - a posture information output unit (14, S130) configured to output information based on a physical quantity estimated by the estimation unit (12, 13, S112 to S118) as posture information indicative of a posture of the vehicle (2);- a reliability determination unit (14, S120, S122) configured to determine a reliability of a gyro sensor (21) provided in the vehicle (2) based on a result of a comparison of a posture of the vehicle (2), indicated by a physical quantity estimated by the estimation unit (12, 13, S112 to S118), with a measurement associated with a posture of the vehicle (2) derived from the gyro sensor (21); and - a reliability information output unit (14, S124) configured to output unreliability information, which is information indicating that the reliability of the gyro sensor (21) is determined to be unreliable when the reliability of the gyro sensor (21) is determined to be unreliable by the reliability determination unit (14, S120, S122); An attitude estimation device (1) comprising:- an image acquisition unit (11) configured to acquire multiple images from multiple cameras (20a to 20d) mounted on a vehicle (2) to capture images around the vehicle (2), wherein the multiple cameras (20a to 20d) are mounted to form multiple overlapping regions (3, 4, 5, 6), wherein the overlapping regions (3, 4, 5, 6) are regions in which image regions partially overlap;- a detection unit (11, 12, S100 to S110) configured to detect at least one feature portion, which is a portion indicating a predetermined feature in an area, on each of multiple images each including appearances of the overlapping regions (3, 4, 5, 6), wherein the multiple images are acquired by the image acquisition unit (11) at approximately the same time;- an estimation unit (12, 13, S112 to S118) configured to calculate, for each of the overlapping areas (3, 4, 5, 6), an amount of positional offset between the feature portions acquired from the plurality of images each including appearances of the overlapping areas (3, 4, 5, 6), and to estimate a predetermined physical quantity associated with a posture of the vehicle (2) based on a difference in the amount of offset calculated for each of the plurality of overlapping areas (3, 4, 5, 6); - a posture information output unit (14, S130) configured to output information based on a physical quantity estimated by the estimation unit (12, 13, S112 to S118) as posture information indicative of a posture of the vehicle (2);and- an environment determination unit (14, S128) that determines a predetermined environment linked to characteristics of the gyro sensor (21) provided in the vehicle (2), wherein- the attitude information output unit (14, S130) is configured to selectively execute control to output information based on a physical quantity estimated by the estimation unit (12, 13) as the attitude information according to a determination made by the environment determination unit (14, S128) or control to output information based on a measurement detected by the gyro sensor (21) as the attitude information.;

Images