Object position calculation device

Abstract

The present invention obtains relative position information of an external object identified by a plurality of identifiers (11, 41) without requiring complicated computation or obtaining self-position accuracy information. An object position calculation device (100) comprises: a first identifier (11) for identifying a fixed object; a first position information acquisition unit (15) for acquiring first relative position information; a second identifier (41) for identifying the fixed object; a second position information acquisition unit (45) for acquiring second relative position information; a first difference calculation unit (21) for calculating the difference value between the second relative position information and the first relative position information; a second difference calculation unit (22) for calculating the difference value between a first position information difference value and a position information correction value; a correction value update unit (23) for updating the position information correction value on the basis of a second position information difference value; and an object position calculation unit (26) for calculating relative position information of the fixed object by correcting the first relative position information on the basis of the updated correction value.

Description

Object position calculation device

[0001] The present disclosure relates to an object position calculation device.

[0002] Conventionally, there has been known a technique for estimating the position of a moving body using information from an object position calculation device mounted on the moving body. The moving body is a movable object with a control function, such as an automobile or a robot. The object position calculation device mounted on the moving body is provided with a classifier for identifying the object. Examples of classifiers include devices that measure information about objects outside the moving body, such as a visible light camera or radar, devices that measure information about the moving body, such as acceleration and angular velocity, and devices that acquire information about the outside of the moving body from data storage, such as a map.

[0003] Furthermore, the position of a moving body can be defined as information that indicates the absolute or relative position of the moving body. The position of a moving body does not necessarily have to be information (such as latitude and longitude) that indicates the absolute position of the moving body on the Earth. Information that indicates the positional relationship between a target object that exists outside the moving body and the vehicle (for example, the relative distance between the vehicle and the target object) is also information that indicates the position of the moving body. Here, a target object is an object whose position can be identified.

[0004] A technique has been disclosed in which the distance to a target outside the moving body is measured relative to the moving body's predicted self-position, and the self-position is corrected using the measured distance information (for example, Patent Document 1).

[0005] Patent No. 6968877

[0006] The technology disclosed in Patent Document 1 acquires self-location accuracy information of a moving object and determines a gain for calculating a correction value based on this self-location accuracy information. Then, a correction value for correcting the self-location is calculated from the difference between the measured distances from the moving object to the target object calculated using two different methods. The self-location accuracy information described in Patent Document 1 is calculated as variances in the X and Y directions, which are obtained by converting a covariance matrix into a body coordinate system using a Jacobian matrix based on the positional relationship between the vehicle's estimated own position and the feature. Performing complex calculations after calculating the self-location in this way places a heavy burden on the object position calculation device, which can cause delays in calculation processing or increase the cost of the processing device.

[0007] The present disclosure aims to solve such problems by providing an object position calculation device that can update a correction value based on the difference between relative position information of external objects identified by multiple classifiers and calculate object position information through appropriate correction, without requiring complex calculations or obtaining self-position accuracy information.

[0008] an object position calculation device according to the present disclosure, the object position calculation device including: a first identifier that identifies a fixed object existing outside the moving body; a first position information acquisition unit that acquires first relative position information between the fixed object identified by the first identifier and the moving body; a second identifier that identifies the fixed object; a second position information acquisition unit that acquires second relative position information between the fixed object identified by the second identifier and the moving body; a first difference calculation unit that calculates a difference value between the second relative position information of the fixed object acquired by the second position information acquisition unit and the first relative position information of the fixed object acquired by the first position information acquisition unit, and outputs the difference value as the first position information difference value; a second difference calculation unit that calculates a difference value between the difference value of the first position information output by the first difference calculation unit and a correction value of previously updated position information, and outputs the difference value as the second position information difference value; a correction value update unit that updates the correction value of the position information based on the difference value of the second position information output by the second difference calculation unit; and an object position calculation unit that calculates the relative position information of the fixed object by correcting the first relative position information based on the correction value of the position information updated by the correction value update unit.

[0009] According to the object position calculation device disclosed herein, it is possible to obtain an object position calculation device that does not require complex calculations and does not require self-position accuracy information, but updates correction values ​​based on differences in relative position information of external objects identified by multiple classifiers, and calculates object position information through appropriate correction. This makes it possible to perform appropriate corrections and calculate object positions with high accuracy while suppressing delays in calculation processing and increases in the cost of the processing device.

[0010] FIG. 1 is a configuration diagram of an object position calculation device according to embodiment 1. FIG. 2 is a diagram illustrating an example of a relative position between a moving body and an object according to embodiment 1. FIG. 3 is a diagram illustrating a hardware configuration diagram of an object position calculation device according to embodiment 1. FIG. 4 is a first flowchart illustrating processing by the object position calculation device according to embodiment 1. FIG. 5 is a second flowchart illustrating processing by the object position calculation device according to embodiment 1. FIG. 6 is a configuration diagram of an object position calculation device according to embodiment 2. FIG. 7 is a second flowchart illustrating processing by the object position calculation device according to embodiment 2. FIG. 8 is a configuration diagram of an object position calculation device according to embodiment 3. FIG. 9 is a first flowchart illustrating processing by the object position calculation device according to embodiment 3. FIG. 10 is a second flowchart illustrating processing by the object position calculation device according to embodiment 3. FIG. 11 is a configuration diagram of an object position calculation device according to embodiment 4. FIG. 12 is a first flowchart illustrating processing by the object position calculation device according to embodiment 4. FIG. 13 is a configuration diagram of an object position calculation device according to embodiment 5. FIG. 14 is a configuration diagram of an object position calculation device according to embodiment 6. FIG. 15 is a diagram illustrating an example of a relative position between a moving body and an object according to embodiment 6. FIG. 16 is a first flowchart illustrating processing by the object position calculation device according to embodiment 6. FIG. 17 is a second flowchart illustrating processing by the object position calculation device according to embodiment 6. FIG. 18 is a configuration diagram of an object position calculation device according to embodiment 7. 10 is a first flowchart showing the processing of an object position calculation device according to Embodiment 7. FIG. 11 is a second flowchart showing the processing of an object position calculation device according to Embodiment 7. FIG.

[0011] Hereinafter, embodiments will be described in detail with reference to the drawings. Note that the drawings are schematic, and for the sake of convenience, configurations may be omitted or simplified as appropriate. In the following description, similar components are denoted by the same reference numerals, and their names and functions are also the same. Therefore, detailed descriptions thereof may be omitted to avoid duplication.

[0012] 1. Embodiment 1 <Configuration of Object Position Calculation Device> Fig. 1 is a configuration diagram of an object position calculation device 100 according to embodiment 1. The object position calculation device 100 mounted on a moving object performs appropriate corrections based on relative position information to a target acquired by a first position information acquisition unit 15 and relative position information to a target acquired by a second position information acquisition unit 45, and outputs highly accurate relative position information of the target.

[0013] A target refers to an object whose position can be identified. Here, a target refers to a fixed object that exists outside a moving object and whose position is defined on a map. Hereinafter, a target will be referred to as an object. A target that is particularly distinctive and easy to identify may be called a landmark. A moving object is a general term for controlled, movable objects such as automobiles, railway vehicles, aircraft, and robots. Hereinafter, a moving object may be described using a vehicle as an example.

[0014] The object position calculation device 100 has the following functional blocks. The first classifier 11 and the second classifier 41 identify relative position information of an object present outside the moving body, and transmit the first relative position information to the first position information acquisition unit 15 and the second relative position information to the second position information acquisition unit 45. The first classifier 11 and the first position information acquisition unit 15 may be collectively referred to as a first recognition device. The second classifier 41 and the second position information acquisition unit 45 may be collectively referred to as a second recognition device.

[0015] A difference value between the second relative position information acquired by the second position information acquisition unit 45 and the first relative position information acquired by the first position information acquisition unit 15 is calculated by the first difference calculation unit 21 and transmitted as a difference value of the first position information to the second difference calculation unit 22. The second difference calculation unit 22 calculates a difference value between the correction value and the first relative position information and transmits it to the correction value update unit 23 as a difference value of the second position information.

[0016] The correction value update unit 23 updates the correction value based on the difference value of the second position information. The updated correction value is stored in the correction value storage unit 25. The object position calculation unit 26 corrects the first relative position information acquired by the first position information acquisition unit 15 using the updated correction value, calculates the relative position of the object, and outputs it as object relative position information. The update necessity determination unit 24 determines whether or not to update the correction value.

[0017] <Classifier> The first classifier 11 and the second classifier 41 identify the relative positions of objects present outside the mobile body. The first classifier 11 and the second classifier 41 are configured by external monitoring sensors that detect objects around the mobile body, or a combination of a positioning device and a map data device.

[0018] External surveillance sensors can be image sensors, radio wave sensors, optical sensors, ultrasonic sensors, etc. Image sensors, such as surveillance cameras, capture images of objects and calculate the distance to the object from the image data captured within a certain viewing angle range. The image data can also be used to obtain the size, direction of movement, speed, and type of the object. Visible light cameras, infrared cameras, etc. can be used as image sensors.

[0019] The radio wave sensor can be a millimeter wave radar (MMWR) that uses a frequency band of 24 to 79 GHz, etc. The radio wave sensor can detect the position of an object and the speed of the object's movement by the Doppler effect.

[0020] Examples of optical sensors that can be used include laser radar and LiDAR (Light Detection and Ranging). LiDAR irradiates a laser beam within a certain field of view and detects point cloud data obtained by the reflection of the laser beam from an object, allowing the position and shape of the object to be determined.

[0021] Information processing may be performed for each sensor that grasps the external environment, such as an image sensor, a radio wave sensor, an optical sensor, an ultrasonic sensor, etc. In this way, data acquired by the various sensors can be processed, and only information about the identified object (for example, relative position, etc.) can be transmitted to the first classifier 11 and the second classifier 41.

[0022] The first classifier 11 and the second classifier 41 may use any of an image sensor, a radio wave sensor, an optical sensor, and an ultrasonic sensor, or may use a combination of information from a plurality of sensors. Furthermore, other sensors may be used to grasp the external environment.

[0023] The positioning device may calculate the vehicle position using positioning information from a Global Navigation Satellite System (GNSS) that detects the vehicle position. The positioning device may calculate the vehicle position using a travel distance sensor that detects the number of rotations of the vehicle's wheels, a gyro sensor that detects the vehicle's acceleration, speed, angular acceleration, and angular velocity, etc. The positioning device may identify the vehicle position by receiving information from a tag or the like embedded in the road using a short-range wireless communication technology such as Near Field Communication (NFC).

[0024] The vehicle position information detected by the positioning device is compared with the position information of the object described in the map data of the map data device, and relative position information between the vehicle and the object is output. The first classifier 11 and the second classifier 41 may acquire the relative position information from the combination of the positioning device and the map data device.

[0025] <Relative Position Information> Information on the relative position of an object may be expressed using X-Y coordinates with the east-west direction as the X axis and the north-south direction as the Y axis, or X-Y coordinates with the front-to-back direction of the moving object as the Y axis and the left-to-right direction as the X axis. Furthermore, since the Earth is a sphere, it may also be expressed using coordinates that use latitude and longitude. Furthermore, it may also be expressed using vectors that indicate the distance and direction from the moving object to the object. Comparison and correction of relative position information will be explained using FIG. 2.

[0026] 2 is a diagram showing an example of the relative positions of a moving body and an object according to embodiment 1. In FIG. 2, the moving body is a vehicle 50, and the relative positions of the object are shown in XY coordinates, with the vehicle 50 as the origin, and the left-right direction as the X axis and the front-rear direction as the Y axis.

[0027] <Calculation of each vector> The first relative position information acquired by the first position information acquisition unit is indicated by a triangle mark as P1 (X1, Y1). In other words, the relative position of the object from the vehicle is indicated by the vector P1 (solid arrow). The second relative position information of the object acquired by the second position information acquisition unit is indicated by a double circle mark as P2 (X2, Y2). In other words, the relative position of the object from the vehicle is indicated by the vector P2 (solid arrow).

[0028] The difference value of the first position information calculated by the first difference calculation unit 21 is indicated by vector D1 (dash-dotted line). Vector D1 is the difference obtained by subtracting first relative position information P1 (X1, Y1) from second relative position information P2 (X2, Y2), i.e., vector P1 subtracted from vector P2. In FIG. 2, the X component of vector D1 is calculated as X2-X1, and the Y component is calculated as Y2-Y1.

[0029] The difference value of the second position information calculated by the second difference calculation unit 22 is indicated by a vector D2 (dashed line) extending from the square mark to the double circle mark. Vector D2 is the difference obtained by subtracting vector CL, which is the latest value of the correction value, from vector D1, which is the difference value of the first position information. Vector CL is indicated in FIG. 2 by a vector (dashed line) extending from the triangle mark to the square mark. If vector D2, which is the second relative position information, is added directly to vector CL, which is the latest correction value before updating, to obtain a new correction value, a correction value for calculating second relative position information P2 based on first relative position information P1 is obtained.

[0030] Here, the pre-update correction value (vector CL) is updated using a gain K. Specifically, the correction value update unit 23 calculates the post-update correction value (vector CN) based on the difference value (vector D2) of the second position information and the pre-update correction value (vector CL). The post-update correction value is calculated, for example, using the following formula:

[0031] (Updated correction value CN) = (Pre-update correction value CL) + K × (Difference value D2 of second position information) (1) Here, K is a gain, which is a parameter that adjusts the level of responsiveness of the correction value to time fluctuations. K is a value between 0 and 1. K may be a fixed value, or may be a variable value based on an existing algorithm such as a Kalman filter.

[0032] Now, when K = 0.4, K × (the difference value of the second position information) is calculated as (K × vector D2) in Figure 2 and is represented as a vector extending from the square mark to the circle mark. The updated correction value is indicated by vector CN in Figure 2 and is represented as a vector extending from the triangle mark to the circle mark.

[0033] The X and Y components of each vector can be defined as follows: First relative position information: Vector P1 (X1, Y1) Second relative position information: Vector P2 (X2, Y2) Difference value of first position information: Vector D1 (XD1, YD1) Difference value of second position information: Vector D2 (XD2, YD2) Correction value before update: Vector CL (XL, YL) Correction value after update: Vector CN (XN, YN) Relative position information of object calculated using the correction value after update: Vector PO (XO, YO)

[0034] The relationships between the vectors are as follows: Vector D1 = Vector P2 - Vector P1 Vector D2 = Vector D1 - Vector CL Vector CN = Vector CL + K x Vector D2 Vector PO = Vector P1 + Vector CN

[0035] <About the X coordinate> XD1=X2-X1 XD2=XD1-XL=X2-X1-XL XN=XL+K×XD2=XL+K×(X2-X1-XL) YD2=YD1-YL=Y2-Y1-YL YN=YL+K×YD2=YL+K×(Y2-Y1-YL) YO=Y1+YN=Y1+YL+K×(Y2-Y1-YL)

[0036] This procedure eliminates the need for complex calculations and allows the correction value to be updated based on the difference in relative position information of external objects identified by two types of classifiers, without the need to obtain self-position accuracy information. This makes it possible to obtain an object position calculation device that can calculate object position information through appropriate correction. This makes it possible to perform appropriate correction and calculate object positions with high accuracy while suppressing delays in calculation processing and increases in the cost of the processing device.

[0037] 2, the first relative position information P1 regarding an object acquired by the first position information acquisition unit 15 is a different value from the second relative position information P2 regarding the same object acquired by the second position information acquisition unit 45. This is because the accuracy of the first classifier 11 and the second classifier 41 is different.

[0038] For example, a sensor that can identify objects over a long distance may be used as the first classifier 11. Then, a sensor that has a short identification distance but has high accuracy in identifying objects over a short distance may be used as the second classifier 41.

[0039] The object position calculation device 100 outputs relative position information calculated by correcting the first relative position information of an object acquired from the first classifier 11 and the second relative position information to the same object acquired from the second classifier 41. By using this corrected relative position information, it becomes possible to utilize both the advantage of the first classifier 11, which can identify objects at distant locations, and the advantage of the second classifier 41, which has high accuracy in identifying relative positions in close locations.

[0040] When the object to be identified is a landmark such as a building or sign, a millimeter-wave radar may be used as the first identifier 11, and a LiDAR or an image sensor may be used as the second identifier 41. Millimeter-wave radar can detect objects at relatively long distances and can identify objects to a certain extent even in environments with fog, mist, or rain. In contrast, LiDAR and image sensors can identify objects with high accuracy at short distances, but the distance at which they can be identified is shortened in environments with fog, mist, or rain.

[0041] The difference in characteristics between the first classifier 11 and the second classifier 41 does not necessarily have to be the same as in the above example. For example, the first classifier 11 may have high accuracy in measuring relative distance, while the second classifier 41 may have high accuracy in measuring relative angle. Even in this case, millimeter-wave radar may be used as the first classifier 11, and LiDAR or an image sensor may be used as the second classifier 41. Millimeter-wave radar can accurately identify the distance to an object, but does not have high accuracy in identifying relative angles. In contrast, LiDAR or an image sensor has high accuracy in identifying relative angles.

[0042] <Hardware Configuration of Object Position Calculation Device> FIG. 3 is a hardware configuration diagram of an object position calculation device. In this embodiment, object position calculation device 100 is an electronic control device mounted on a mobile body to calculate relative position information of an object located outside the mobile body. Each function of object position calculation device 100 is realized by a processing circuit provided in object position calculation device 100. Specifically, object position calculation device 100 includes, as processing circuits, an arithmetic processing device 90 (computer) such as a CPU (Central Processing Unit), a storage device 91 that exchanges data with the arithmetic processing device 90, an input circuit 92 that inputs external signals to the arithmetic processing device 90, and an output circuit 93 that outputs signals from the arithmetic processing device 90 to the outside. Each piece of hardware, such as the arithmetic processing device 90, the storage device 91, the input circuit 92, and the output circuit 93, is connected to one another via a wired or wireless network such as a bus.

[0043] The arithmetic processing device 90 may include an application-specific integrated circuit (ASIC), an integrated circuit (IC), a digital signal processor (DSP), a graphics processing unit (GPU), a field programmable gate array (FPGA), various logic circuits, various signal processing circuits, etc. Furthermore, the arithmetic processing device 90 may include a plurality of the same or different types of devices, each performing a different process. The storage device 91 may include a random access memory (RAM) configured to be able to read and write data from the arithmetic processing device 90, a read-only memory (ROM) configured to be able to read data from the arithmetic processing device 90, etc. The storage device 91 may include non-volatile or volatile semiconductor memory such as a flash memory, a solid-state drive (SSD), an EPROM, an EEPROM, a magnetic disk, a flexible disk, an optical disk, a compact disk, a minidisk, a DVD, etc. The input circuit 92 is connected to various sensors, switches, and communication lines, and includes an A / D converter and a communication circuit that input output signals and communication information from these sensors and switches to the arithmetic processing device 90. The output circuit 93 includes a drive circuit and a communication circuit that output control signals from the arithmetic processing device 90. The interfaces of the input circuit 92 and the output circuit 93 may be based on specifications such as CAN (Control Area Network) (registered trademark), Ethernet (registered trademark), USB (Universal Serial Bus) (registered trademark), DVI (Digital Visual Interface) (registered trademark), and HDMI (High-Definition Multimedia Interface) (registered trademark). Furthermore, the arithmetic processing device 90 may be directly connected to a communication device separate from the input circuit 92 and the output circuit 93 to perform communication.

[0044] Each function of the object position calculation device 100 is realized by the arithmetic processing device 90 executing software (programs) stored in a storage device 91 such as a ROM, and cooperating with other hardware of the object position calculation device 100, such as the storage device 91, input circuitry 92, and output circuitry 93. Setting data such as thresholds and judgment values ​​used by the object position calculation device 100 is stored in the storage device 91 such as a ROM as part of the software (program). Each function of the object position calculation device 100 may be configured as a software module, or may be configured as a combination of software and hardware.

[0045] <Processing of Object Position Calculation Device> Fig. 4 is a first flowchart showing the processing of the object position calculation device 100 according to embodiment 1. Fig. 5 is a second flowchart showing the processing of the object position calculation device 100. Fig. 5 shows a continuation of the processing of Fig. 4.

[0046] 4 and 5 are executed by the arithmetic processing unit of the object position calculation device 100. This processing may be executed at predetermined time intervals (for example, every 1 ms). Instead of at predetermined time intervals, this processing may be executed in response to an event such as the first classifier 11 and the second classifier 41 of the object position calculation device 100 identifying an object.

[0047] Consider a case where the discrimination period in which the first classifier 11 and the second classifier 41 identify external objects differs from the period in which the object position calculation device 100 periodically updates the correction values. The relative position information to the object used by the object position calculation device 100 is the relative position information to the object in the most recent discrimination period. Here, the most recent discrimination period can be defined as the past discrimination period closest to the current time. Note that the relative position information used by the object position calculation device 100 does not necessarily have to be the relative position information to the object in the most recent discrimination period. Relative position information from the moving body to the object in a discrimination period older than the current time, or an average value thereof, may also be used. The relative position information of the external objects identified by the first classifier 11 and the second classifier 41 may be subjected to appropriate filtering to remove noise, and the first position information acquisition unit 15 and the second position information acquisition unit 45 may acquire the information.

[0048] 4 starts, and in step S101, the first position information acquisition unit 15 acquires first relative position information P1 of the object identified by the first classifier 11. Note that if the first position information acquisition unit 15 fails to identify the first relative position information P1 of the object, for example, if there is no object that can be measured by the first classifier 11, or if the measurement device of the first classifier 11 has stopped functioning, the first relative position information P1 is determined to be an invalid value and stored.

[0049] In step S102, the second position information acquisition unit 45 acquires second relative position information P2 of the object identified by the second classifier 41. If the second position information acquisition unit 45 fails to identify the second relative position information P2 of the object, for example, if there is no object that can be measured by the second classifier 41, or if the measurement device of the second classifier 41 has stopped functioning, the second relative position information P2 is determined to be an invalid value and stored.

[0050] In step S103, the first difference calculation unit 21 calculates a value obtained by subtracting the first relative position information P1 from the second relative position information P2 as a difference value D1 of the first position information. The difference value may be calculated using the X-axis coordinate component and the Y-axis coordinate component of each piece of position information. If either the first relative position information P1 or the second relative position information P2 is an invalid value, the first position information difference value D1 is determined to be an invalid value and stored.

[0051] In step S104, the latest correction value stored in the correction value storage unit 25 is read as a pre-update correction value CL. If a pre-update correction value CL is not stored in the correction value storage unit 25 at the time step S104 is executed, an initial value of the correction value is set as the pre-update correction value CL. This initial value may be a predetermined fixed value, or may be set as a value that changes depending on the surrounding environment, etc.

[0052] In step S105, the second difference calculation unit 22 calculates a value obtained by subtracting the pre-update correction value CL from the difference value D1 of the first position information as the difference value D2 of the second position information. However, if the difference value D1 of the first position information is an invalid value, the difference value D2 of the second position information is also determined to be an invalid value and stored.

[0053] In step S106, it is confirmed that the difference value D1 of the first location information or the difference value D2 of the second location information is both valid. Then, in step S107, it is determined whether both difference values ​​are valid. If both are valid (determined YES), the process proceeds to step S112. If either difference value is not valid (determined NO), the process proceeds to step S115.

[0054] In step S112, the correction value is updated. Specifically, the correction value update unit 23 calculates an updated correction value CN based on the difference value D2 of the second position information and the pre-update correction value CL. The updated correction value CN is calculated using equation (1).

[0055] The correction value may be updated by a method other than the above in step S112. In step S113, the correction value update unit 23 stores the calculated updated correction value CN in the correction value storage unit 25.

[0056] In step S115, the updated correction value CN is added to the first relative position information P1 acquired by the object position calculation unit 26. The added value is calculated as corrected relative position information PO. Then, the process ends.

[0057] In the above description, the difference value D1 of the first position information is calculated as (second relative position information P2) - (first relative position information P1) in step S103 (i.e., the value obtained by subtracting the first relative position information P1 from the second relative position information P2 is used). Alternatively, in step S103, the difference value D1 of the first position information may be calculated as (first relative position information P1) - (second relative position information P2). In that case, however, it is necessary to calculate the difference value D2 of the second position information as (difference value D1 of the first position information) + (pre-update correction value CL) in step S105.

[0058] Alternatively, the correction value can be updated in step S112 based on the following formula:

[0059] (Updated correction value CN)=K×(difference value D1 of first position information)+(1−K)×(pre-updated correction value CL) (2) Here, K is a parameter that adjusts the level of responsiveness of the correction value to time fluctuations, and is a value between 0 and 1. The correction value may be updated in step S112 using a method other than the above.

[0060] According to the object position calculation device 100 of the first embodiment, the updated correction value CN calculated by the correction value update unit 23 is stored in the correction value storage unit 25, and then the object position calculation unit 26 can correct the relative position information to the object using the stored correction value. Furthermore, if the difference value D1 of the first position information or the difference value D2 of the second position information is invalid, the correction value stored in the correction value storage unit 25 is sent to the object position calculation unit 26 without being changed. With this configuration, even if the first position information acquisition unit 15 or the second position information acquisition unit 45 fails to acquire the relative distance to the object, the first relative position information is corrected using the correction value calculated in a past period, and the relative position information is calculated.

[0061] 2 described above, assume that the second classifier 41 is a visible light camera. With the configuration of the first embodiment, even if the camera fails to measure the lane marking due to a temporary change in the amount of light, it is possible to correct an error in the relative distance to the lane marking with respect to the first classifier 11. In other words, even if the classifier temporarily cannot effectively identify an object, it is possible to reduce the frequency of an event in which the estimation accuracy of the relative position information to the object and the vehicle position decreases.

[0062] 2. Second Embodiment <Configuration of Object Position Calculation Device> Fig. 6 is a configuration diagram of an object position calculation device 100 according to a second embodiment. Fig. 6 differs from the configuration diagram of Fig. 1 according to the first embodiment in that a second classifier accuracy information acquisition unit 32 and a map database 33 are added. The hardware configuration diagram of Fig. 3 can also be applied to the second embodiment.

[0063] In embodiment 1, if the relative position information to the object output from the first position information acquisition unit 15 and the second position information acquisition unit 45 is valid, the correction value is updated and the relative position information is corrected using the updated correction value, and if the relative position information to the object is not valid, the correction value is not updated and the relative position information is corrected.

[0064] However, even if the actual position information acquisition unit acquires valid relative position information, the reliability of the relative position information may be low. In this case, in the first embodiment, the correction value is updated using the relative position information with low reliability, and the updated correction value is sent to the object position calculation unit 26. Then, there is a possibility that the accuracy of the relative position information to the object corrected using the updated correction value will be degraded.

[0065] <Classifier Accuracy Information> Therefore, the object position calculation device 100 according to the second embodiment is provided with a second classifier accuracy information acquisition unit 32 that acquires classification accuracy information of the second classifier 41. The update gain updated by the correction value update unit 23 can be changed according to the second classifier accuracy acquired by the second classifier accuracy information acquisition unit 32, and the update necessity determination unit 24 can determine whether an update is necessary. With this configuration, when the accuracy of the second relative position information acquired by the second position information acquisition unit 45 is low, it is possible to reduce the frequency of accuracy degradation of the corrected relative position information to the object.

[0066] Note that the identification accuracy information of the second classifier 41 is defined here as information related to the measurement accuracy of the relative position information acquired from the second classifier 41. For example, when the second relative position information of an object is identified by the second classifier 41, the identification accuracy information of the second classifier 41 may be, for example, a standard deviation related to the measurement error of the relative distance. The identification accuracy information of the second classifier 41 may also be information representing conditions under which the measurement error of the second position information acquisition unit 45 increases or decreases. For example, when the second position information acquisition unit 45 is a visible light camera with an image sensor, the identification accuracy information of the second classifier 41 may also include information on whether the visible light camera is capturing an area with a high amount of light or an area with a low amount of light, information on dirt adhering to the imaging lens of the visible light camera, information representing the magnitude of vibration relative to the visible light camera, and the like.

[0067] Furthermore, the classification accuracy information of the second classifier 41 may be set for each road section. The classification accuracy information for each road section may be written in the map data stored in the map database 33. In this case, the second classifier accuracy information acquisition unit 32 can acquire the classification accuracy information of the second classifier 41 by exchanging information with the map database 33.

[0068] If the second classifier 41 is a visible light camera with an image sensor, classification accuracy will decrease in road sections inside tunnels and road sections that include tunnel exits. A decrease in the amount of ambient light inside a tunnel or a sudden change in the amount of light at the tunnel exit area can cause errors in the recognition of images captured by the visible light camera, leading to a decrease in accuracy. The second classifier accuracy information acquisition unit 32 may acquire the expected degree of accuracy decrease, thereby changing the gain for updating the correction value in this road section to slow down the update speed. The gain for updating the correction value in a road section may be changed by setting stages for the degree of accuracy decrease. Furthermore, updating the correction value may be prohibited in road sections inside tunnels and including tunnel exits.

[0069] In this way, road sections where the identification accuracy decreases may be registered in the data of the map database 33. As road sections where the identification accuracy decreases, the degree of decrease in identification accuracy may be registered for roads that run through forests with poor visibility, road sections where fog, mist or rain occurs, road sections where nighttime lighting facilities are insufficient, etc.

[0070] <Processing of object position calculation device> Fig. 7 is a second flowchart showing the processing of object position calculation device 100 according to embodiment 2. The first flowchart showing the processing of object position calculation device 100 according to embodiment 2 is the same as Fig. 4. Fig. 7 shows a continuation of the processing of Fig. 4.

[0071] The flowchart in Fig. 7 differs from the flowchart in Fig. 5 in that steps S108 to S111 are added between steps S107 and S112. The differences will be mainly described below.

[0072] 4 and 7 are executed by the arithmetic processing unit of the object position calculation device 100. This processing may be executed at predetermined time intervals (for example, every 1 ms). Instead of at predetermined time intervals, this processing may be executed in response to an event such as the first classifier 11 and the second classifier 41 of the object position calculation device 100 identifying an object.

[0073] After the process of Fig. 4, in step S107 of Fig. 7, it is determined whether both difference values ​​are valid. If both are valid (determined YES), the process proceeds to step S108. If either difference value is not valid (determined NO), the process proceeds to step S115.

[0074] In step S108, the second classifier accuracy information acquisition unit 32 acquires the classification accuracy information of the second classifier 41. In step S109, it is determined whether the accuracy of the acquired classification accuracy information is greater than a predetermined classification accuracy judgment value. If the accuracy is greater than the classification accuracy judgment value (determination is YES), the process proceeds to step S110. If the accuracy is equal to or less than the classification accuracy judgment value (determination is NO), the process proceeds to step S111.

[0075] In step S110, a gain K for updating the correction value is set. Since the accuracy is greater than the discrimination accuracy judgment value, a high gain value KH (for example, 0.6) is set as the gain K. Then, the process proceeds to step S112.

[0076] In step S111, a gain K (gain K) for updating the correction value is set. Since the accuracy is equal to or lower than the discrimination accuracy judgment value, a low gain value KL (e.g., 0.2) is set as the gain K. Then, the process proceeds to step S112. Here, if the gain K is set to 0, no correction is performed and the pre-update correction value is used as is.

[0077] In step S112, the correction value is updated. The correction value update unit 23 calculates an updated correction value CN based on the difference value D2 of the second position information and the pre-update correction value CL using the gain K for updating the correction value. The updated correction value CN is calculated using equation (1).

[0078] According to the object position calculation device 100 according to the second embodiment configured as described above, the second classifier accuracy information acquisition unit 32 acquires the classification accuracy information of the second classifier 41, thereby making it possible to update the correction value according to the classification accuracy of the second classifier 41. As a result, when the relative position information output from the second position information acquisition unit 45 is a valid value but its accuracy is low, the gain K according to the accuracy is set and the correction value is updated.

[0079] Therefore, the update speed of the correction value is adjusted according to the change in the accuracy of the second classifier 41. That is, the degree of reflection of the second relative position information identified by the second classifier 41 is adjusted. Therefore, it is possible to limit the deterioration in the accuracy of the corrected relative position information.

[0080] Furthermore, by acquiring the identification accuracy information specified for each road section, it is possible to appropriately acquire the identification accuracy information of the second classifier 41 without performing complex and large-scale calculations to obtain self-location accuracy information. Therefore, it is possible to perform appropriate corrections and calculate the position of an object with high accuracy while suppressing delays in calculation processing and increases in the cost of the processing device.

[0081] 3. Third Embodiment <Configuration of Object Position Calculation Device> Fig. 8 is a configuration diagram of an object position calculation device 100 according to a third embodiment. The configuration diagram of Fig. 8 differs from the configuration diagram of Fig. 1 according to the first embodiment in that a correction value prediction unit 34 is added. The hardware configuration diagram of Fig. 3 can also be applied to the third embodiment.

[0082] The appropriate correction value used in the object position calculation device 100 may fluctuate over time. For example, consider a case where the first classifier 11 identifies relative position information from a vehicle to an object using a combination of a positioning device that performs inertial navigation using a gyro sensor and a map data device. In general, errors in inertial navigation accumulate and increase over time or travel distance. In such a case, the difference between the first classifier 11 and the second relative position information obtained by the second classifier 41, which identifies relative position information using another method, tends to increase.

[0083] If, for some reason, the second classifier 41 cannot obtain relative position information, the object position calculation device 100 according to embodiment 1 corrects the first relative position information using the correction value before updating and outputs the corrected first relative position information as object relative position information. If the state in which the second classifier 41 cannot obtain relative position information continues, the same correction value before updating continues to be used.

[0084] <Correction Value Prediction Unit> If the correction value changes over time, it is not appropriate to continue using the same pre-update correction value. This is because continuing to use the same correction value will change the error in the corrected relative position information to the object. This error increases the longer the time continues when the second classifier 41 cannot obtain relative position information.

[0085] In the object position calculation device 100 according to the third embodiment, future correction values ​​are predicted by the correction value prediction unit 34. The correction value prediction unit 34 predicts future correction values ​​based on a history of past correction value updates stored in the correction value storage unit 25. The correction value prediction unit 34 calculates a correction value that is appropriate at the current time and stores it in the correction value storage unit 25. With this configuration, it is possible to predict an appropriate correction value and use the predicted correction value, thereby reducing the frequency of accuracy degradation of the corrected relative position information.

[0086] <Processing of Object Position Calculation Device> Fig. 9 is a first flowchart showing the processing of the object position calculation device 100 according to embodiment 3. Fig. 10 is a second flowchart showing the processing of the object position calculation device 100 according to embodiment 3. Fig. 10 shows a continuation of the processing of Fig. 9.

[0087] The flowchart in Fig. 9 differs from the flowchart in Fig. 4 in that steps S121 to S123 are added before step S101. The flowchart in Fig. 10 differs from the flowchart in Fig. 5 in that steps S124 and S114 are added between steps S107 and S115, and the process proceeds to step S114 after step S113. The main differences will be described below.

[0088] 9 and 10 are executed by the arithmetic processing unit of the object position calculation device 100. This processing may be executed at predetermined time intervals (for example, every 1 ms). Instead of at predetermined time intervals, this processing may be executed in response to an event such as the first classifier 11 and the second classifier 41 of the object position calculation device 100 identifying an object.

[0089] 9 starts, and in step S121 the correction value prediction unit 34 acquires past correction values ​​stored in the correction value storage unit 25. This is to check the history of correction value updates.

[0090] In step S122, the correction value prediction unit 34 predicts a future correction update value based on the history of correction value updates, and sets the predicted correction value as the predicted correction value. Here, the predicted correction value is defined as a correction value at the current time calculated based on a model (prediction model) in which past correction values ​​stored in the correction value storage unit 25 vary. As an example of the prediction model, for example, a model in which the correction value varies linearly with elapsed time T may be used. In this case, the predicted correction value CF may be calculated by the following equation.

[0091] (Predicted correction value CF)=(Pre-update correction value CL)+(Elapsed time T)×(Coefficient KC) (3)Here, the elapsed time T is defined as the time elapsed from the past time (past cycle) when the latest pre-update correction value CL was stored to the current time (current cycle). The coefficient KC may be a predetermined fixed value, or may be a variable value that is set according to the elapsed time using an existing algorithm such as a Kalman filter.

[0092] If the previous correction value acquired in step S121 is invalid, the previous correction value is set as the initial value of the correction value. This initial value may be a predetermined fixed value or may be a variable value set depending on the surrounding environment, etc.

[0093] Then, in step S123, the correction value prediction unit 34 stores the predicted correction value CF in the correction value storage unit 25. Note that if no correction value is stored in the correction value storage unit 25 at the time step S121 is executed, the predicted correction value CF is stored as an invalid value. Then, the process proceeds to step S101, where the same process as in FIG. 4 is executed.

[0094] In step S107 of Fig. 10, it is determined whether both difference values ​​are valid. If both are valid (determination is YES), the process proceeds to step S112. If either difference value is not valid (determination is NO), the process proceeds to step S124.

[0095] In step S124, the predicted correction value CF calculated in step S122 is used as the updated correction value CN to be used in the correction calculation, and is stored in the correction value storage unit 25. Then, the process proceeds to step S114.

[0096] In step S112, the correction value is updated. Specifically, the correction value update unit 23 calculates an updated correction value CN based on the difference value D2 of the second position information and the pre-update correction value CL. The updated correction value CN is calculated using equation (1). In step S113, the correction value update unit 23 stores the calculated updated correction value CN in the correction value storage unit 25. Then, the process proceeds to step S114.

[0097] In step S114, the updated correction value CN is read from the correction value storage unit 25. This is to be used for calculating the corrected relative position information PO.

[0098] Then, in step S115, the object position calculation unit 26 adds the updated correction value CN to the first relative position information P1 acquired. The added value is calculated as corrected relative position information PO. Then, the process ends.

[0099] According to the object position calculation device 100 of the third embodiment, the correction value prediction unit 34 calculates the predicted correction value CF from the update history of past correction values ​​stored in the correction value storage unit 25. If the difference value D1 of the first position information or the difference value D2 of the second position information is invalid, the predicted correction value CF is used as the updated correction value CN instead of using the pre-update correction value.

[0100] The object position calculation unit 26 uses the predicted correction value CF as the updated correction value CN to correct the first relative distance. This configuration allows for appropriate correction even when the valid correction value fluctuates over time and the difference value of the second position information becomes invalid. Because correction is performed using the predicted correction value CF calculated by the correction value prediction unit 34, it is possible to reduce the frequency with which the accuracy of the corrected relative position information deteriorates.

[0101] 4. Fourth Embodiment <Configuration of Object Position Calculation Device> Fig. 11 is a configuration diagram of an object position calculation device 100 according to a fourth embodiment. Fig. 11 differs from the configuration diagram of Fig. 8 according to the third embodiment in that a first classifier accuracy information acquisition unit 31 is added. The hardware configuration diagram of Fig. 3 can also be applied to the fourth embodiment.

[0102] In the object position calculation device 100 according to the third embodiment, the correction value prediction unit 34 predicts future correction values ​​based on a prediction model from the update history of past correction values ​​stored in the correction value storage unit 25. However, a specific prediction model is not always valid.

[0103] For example, consider a case where the first identifier 11 is a device that measures the case where the relative position information from the vehicle to an object is identified by a combination of a positioning device that performs inertial navigation using a gyro sensor and a map data device. The rate at which the error in the inertial navigation increases over time differs between a period when the vehicle is traveling straight and a period when the vehicle is meandering.

[0104] In such a situation, if the correction value prediction unit 34 uses a prediction model that assumes that the vehicle is traveling straight, the predicted correction value will become invalid more frequently during the period when the vehicle is meandering, which increases the frequency with which the accuracy of the corrected relative position information will deteriorate.

[0105] <Classifier Accuracy Acquisition Unit> Therefore, the object position calculation device 100 according to the fourth embodiment adds a first classifier accuracy information acquisition unit 31 that acquires classification accuracy information of the first classifier 11. The correction value prediction unit 34 calculates the predicted correction value CF based on the classification accuracy information of the first classifier 11. With this configuration, it is possible to improve the reliability of the predicted correction value CF even when the appropriate correction value fluctuates over time due to an error in the first position information acquisition unit 15. This reduces the frequency with which the predicted correction value CF significantly deviates from an appropriate value, and as a result, it is possible to reduce the frequency with which the accuracy of the corrected relative position information deteriorates.

[0106] Note that here, the classification accuracy information of the first classifier 11 is defined as information related to the measurement accuracy of the relative position information classified by the first classifier 11 and acquired by the first position information acquisition unit 15. The classification accuracy information of the first classifier 11 may be, for example, the standard deviation of the measurement error of this relative position information.

[0107] Furthermore, the identification accuracy information of the first classifier 11 may be information representing conditions under which the measurement error of the first classifier 11 increases or decreases. For example, if the first classifier 11 is a device that measures the relative distance from the vehicle to an object using GNSS and a map data device, information regarding communication quality with an artificial satellite or the like may be the identification accuracy information of the first classifier 11. Furthermore, if the first classifier 11 is configured by combining a positioning device using a gyro sensor and a map data device and is a device that measures relative position information from the vehicle to an object, information regarding whether the vehicle is traveling in a manner that makes it easy for the error of the gyro sensor to increase may be the identification accuracy information of the first classifier 11.

[0108] <Processing of object position calculation device> Fig. 12 is a first flowchart showing the processing of the object position calculation device 100 according to embodiment 4. The second flowchart showing the processing of the object position calculation device 100 according to embodiment 4 is the same as Fig. 10. Fig. 10 shows a continuation of the processing of Fig. 12.

[0109] The flowchart of Fig. 12 differs from the flowchart of Fig. 9 in that steps S131 to S134 are provided instead of step S122. The differences will be mainly described below.

[0110] 12 and 10 are executed by the arithmetic processing unit of the object position calculation device 100. This processing may be executed at predetermined time intervals (for example, every 1 ms). Instead of at predetermined time intervals, this processing may be executed in response to an event such as the first classifier 11 and the second classifier 41 of the object position calculation device 100 identifying an object.

[0111] 12 starts, and in step S121 the correction value prediction unit 34 acquires past correction values ​​stored in the correction value storage unit 25. This is to check the history of correction value updates.

[0112] In step S131, the first classifier accuracy information acquisition unit 31 acquires the classification accuracy information of the first classifier 11. The classification accuracy information depends on the type of classifier.

[0113] For example, if the first identifier 11 is a device that measures the relative distance from the vehicle to an object using GNSS and a map data device, it can determine that the accuracy is low if the vehicle is blocked by a tunnel or the like and the quality of the communication signal with an artificial satellite or the like is low. Also, for example, if the first identifier 11 is a device configured by combining a positioning device that uses a gyro sensor and a map data device, it can also determine that the accuracy is low if the turning angle of the vehicle is larger than a predetermined threshold.

[0114] In step S132, it is determined whether the accuracy of the acquired identification accuracy information is greater than a predetermined identification accuracy judgment value. If the accuracy is greater than the identification accuracy judgment value (determination is YES), proceed to step S133. If the accuracy is equal to or less than the identification accuracy judgment value (determination is NO), proceed to step S134.

[0115] In step S133, the correction value prediction unit 34 calculates a predicted correction value CF based on a prediction model in which the fluctuation of the correction value is relatively small and on past correction values. Here, the prediction model in which the fluctuation of the correction value is relatively small is defined as a model that is premised on a smaller fluctuation of the correction value per unit time than the prediction model of the correction value used in step S134.

[0116] For example, this model applies when the coefficient KC in equation (3) used in the description of step S122 in embodiment 3 is set to a small value. Also, for example, this model applies when a correction value is predicted using a Kalman filter, and a parameter representing the magnitude of the prediction error is set to a small value. After calculating the predicted correction value CF, the process proceeds to step S123.

[0117] In step S134, the correction value prediction unit 34 calculates a predicted correction value CF based on a prediction model in which the correction value fluctuates relatively greatly and on past correction values. Here, the prediction model in which the correction value fluctuates relatively greatly is defined as a model that is premised on a larger fluctuation in the correction value per unit time than the prediction model of the correction value used in step S133.

[0118] For example, this model applies when the coefficient KC in equation (3) used in the description of step S122 in embodiment 3 is set to a large value. Also, for example, this model applies when a correction value is predicted using a Kalman filter, and a parameter representing the magnitude of the prediction error is set to a large value. After calculating the predicted correction value CF, the process proceeds to step S123.

[0119] In step S123, the correction value prediction unit 34 stores the predicted correction value CF in the correction value storage unit 25. After that, the same process as step S101 in FIG.

[0120] According to the object position calculation device 100 according to the fourth embodiment, the correction value prediction unit 34 is configured to calculate the predicted correction value using a different prediction model depending on the classification accuracy information of the first classifier 11. With this configuration, the frequency with which the predicted correction value significantly deviates from an appropriate value due to time fluctuations in the appropriate correction value caused by errors in the first position information acquisition unit 15 decreases, and as a result, it is possible to reduce the frequency with which the accuracy of the corrected relative distance deteriorates.

[0121] 5. Fifth Embodiment <Configuration of Object Position Calculation Device> Fig. 13 is a configuration diagram of an object position calculation device 100 according to a fifth embodiment. Compared to the configuration diagram of Fig. 1 according to the first embodiment, Fig. 13 clearly indicates that the first identifier 11 is composed of an inertial sensor / speedometer 12, a satellite positioning device 13, and a map database 14. The inertial sensor / speedometer 12 corresponds to a gyro sensor. The satellite positioning device 13 is a device that performs positioning using positioning information from GNSS. The map database 14 corresponds to a map data device. The hardware configuration diagram of Fig. 3 can also be applied to the fourth embodiment.

[0122] In the fifth embodiment, the first position information acquisition unit 15 acquires first relative position information based on information from the inertial sensor / speedometer 12, the satellite positioning device 13, and the map database 14. This solves the problems that arise when an external monitoring sensor such as a visible light camera is used as the first identifier 11. For example, it is possible to reduce the frequency of failure to acquire relative position information due to factors in the measurement environment, such as a decrease in the amount of light around an object or an obstruction that blocks the view of the object.

[0123] The inertial sensor / speedometer 12 acquires data from the inertial sensor and the speedometer in the current cycle. The satellite positioning device 13 acquires satellite positioning information in the current cycle. The map database 14 shares map data within the first classifier 11. The first classifier 11 as a whole compares vehicle position information detected by the positioning device with object position information described in the map data of the map data device, and outputs relative position information between the vehicle and the object to the first position information acquisition unit 15. The first position information acquisition unit 15 transmits the acquired first relative position information to the first difference calculation unit 21 and the object position calculation unit 26.

[0124] The first identifier 11 uses existing technologies such as matching satellite positioning information with map data and an autonomous navigation algorithm using an inertial sensor and a speedometer. If the first identifier 11 fails to identify the relative position information to the object, the first identifier 11 stores the first relative position information as an invalid value. For example, if the satellite positioning information is invalid, if the inertial sensor or the speedometer has stopped functioning, or if the vehicle is traveling in an area where map data does not exist, the first identifier 11 determines that the first relative position information is an invalid value.

[0125] According to the object position calculation device 100 of the fifth embodiment, the first identifier 11 is composed of the inertial sensor / speedometer 12, the satellite positioning device 13, and the map database 14. Therefore, compared to the case where an external monitoring sensor such as an image sensor, a radio wave sensor, or an optical sensor is used, it is possible to reduce the frequency of failure to acquire relative position information due to factors in the measurement environment, such as a decrease in the amount of light around the object or the presence of an obstruction.

[0126] 6. Sixth Embodiment <Configuration of Object Position Calculation Device> Fig. 14 is a configuration diagram of an object position calculation device 100 according to the sixth embodiment. The configuration diagram of Fig. 1 according to the first embodiment differs in that the output in Fig. 14 is specified as the relative distance between a moving body and a fixed object. Since the relative distance between a vehicle and an object is an extremely important factor in vehicle operation, it is important to obtain a highly accurate relative distance by correction. The hardware configuration diagram of Fig. 3 can also be applied to the sixth embodiment.

[0127] <Distance from lane markings> FIG. 15 is a diagram showing an example of the relative position of an object according to the sixth embodiment. For example, the diagram shows a case where the moving body is a vehicle 50 and the object is a lane marking that indicates the boundary between driving lanes on a highway. The vehicle 50 and the first relative position information identified by the first classifier 11 of the object position calculation device 100 mounted on the vehicle 50 and acquired by the first position information acquisition unit 15 are indicated by a solid line 51. The second relative position information identified by the second classifier 41 and acquired by the second position information acquisition unit 45 are indicated by a dashed line 52. The first relative distance L1 and the second relative distance L2 between the vehicle 50 and the object are indicated on the left side of the vehicle 50. The difference ΔL between the first relative distance L1 and the second relative distance L2 is also shown.

[0128] The relative distance to the object can be defined as the distance from a representative point of the moving body equipped with the first classifier 11 and the second classifier 41 to a representative point of the object. The relative distance to the object can be specified as the distance from the center of gravity of the vehicle 50 equipped with the classifier to the lane marking directly to the side of the vehicle. Note that the definition of the relative distance between the object and the vehicle 50 is not necessarily limited to the above example.

[0129] The object position calculation device 100 outputs the corrected relative distance to the object. The corrected relative distance to the object is a relative distance that is corrected by a correction value with respect to the relative distance to the object identified by the first classifier 11 and the second classifier 41.

[0130] The relative distance to the object that the object position calculation device 100 periodically performs a correction process and outputs is the relative distance to the object in the most recent cycle. Here, the most recent cycle can be defined as the most recent cycle in which the process is executed that is closest to the current time. Note that the input and output of the object position calculation device 100 do not necessarily have to be the relative distance to the object in the most recent cycle, and the input and output may also be the relative distance from the vehicle to the object in cycles prior to the current, or the average value of those distances.

[0131] In the example of Figure 15, the first position information acquisition unit 15 is able to acquire relative position information regarding the lane line shape up to positions far from the vehicle. On the other hand, the second position information acquisition unit 45 is only able to acquire relative position information regarding the lane line shape up to positions closer than the first position information acquisition unit 15. Furthermore, the first relative position information from the first classifier 11 is assumed to have lower calculation accuracy for the lane line shape at closer positions than the second relative position information from the second classifier 41. Therefore, the first relative distance L1 to the lane line based on the output of the first position information acquisition unit 15 and the second relative distance L2 to the lane line based on the output of the second position information acquisition unit 45 are shown to be different values.

[0132] The object position calculation device 100 outputs a relative distance calculated by correcting the first relative distance L1 to the lane line, based on the first relative distance L1 to the lane line acquired from the first position information acquisition unit 15 and the second relative distance L2 to the lane line output from the second position information acquisition unit 45. By using this corrected relative distance, it is possible to calculate the relative position of the lane line shape, taking advantage of both the first classifier 11, which can acquire the lane line shape up to distant positions, and the second classifier 41, which has high accuracy in identifying relative positions at close positions.

[0133] The difference in characteristics between the first classifier 11 and the second classifier 41 does not necessarily have to be the same as the above example. For example, the first classifier 11 may have high accuracy in measuring relative distance, while the second classifier 41 may have high accuracy in measuring relative angle.

[0134] <Processing of Object Position Calculation Device> Fig. 16 is a first flowchart showing the processing of the object position calculation device 100 according to embodiment 6. Fig. 17 is a second flowchart showing the processing of the object position calculation device 100. Fig. 17 shows a continuation of the processing of Fig. 16.

[0135] 16 and 17 are executed by the arithmetic processing unit of the object position calculation device 100. This processing may be executed at predetermined time intervals (for example, every 1 ms). Instead of at predetermined time intervals, this processing may be executed in response to an event such as the first classifier 11 and the second classifier 41 of the object position calculation device 100 identifying an object.

[0136] 16 and 17 are obtained by replacing the relative position information in the processes of FIGS. 4 and 5 according to embodiment 1 with relative distances. The processes from step S201 to step S215 are similar to the processes of step S101 to step S115, and therefore will not be described here.

[0137] In the object position calculation device 100 according to the sixth embodiment, the updated correction value calculated by the correction value update unit 23 is stored in the correction value storage unit 25, and then the object position calculation unit 26 corrects the relative distance to the object using the stored correction value. Furthermore, if the difference value of the first position information is an invalid value, the correction value stored in the correction value storage unit 25 is sent to the object position calculation unit 26 without being changed. With this configuration, even if the second position information acquisition unit 45 fails to acquire the relative distance to the object, the corrected relative distance is calculated using the correction value calculated in a past cycle.

[0138] 15 , the object is a lane marking and the second classifier 41 is a visible light camera. With the configuration of the sixth embodiment, even if the camera fails to measure the lane marking due to a temporary decrease in the amount of light, it is possible to correct an error in the relative distance to the lane marking identified by the first classifier 11.

[0139] In other words, with the configuration of embodiment 6, even when the classifier is temporarily unable to measure valid relative position information, it is possible to reduce the frequency of the situation in which the estimation accuracy of the relative distance to the object, i.e., the vehicle position, decreases.

[0140] 7. Seventh Embodiment <Configuration of Object Position Calculation Device> Fig. 18 is a configuration diagram of an object position calculation device 100 according to the seventh embodiment. Compared to the configuration diagram of Fig. 1 according to the first embodiment, Fig. 7 differs in that the output is specified as the relative angle between a moving body and a fixed object. Since the relative angle between a vehicle and an object is an extremely important factor in vehicle operation, it is important to obtain a highly accurate relative angle by correction. The hardware configuration diagram of Fig. 3 can also be applied to the seventh embodiment.

[0141] In the object position calculation device 100 according to the first embodiment, information on the relative position from the vehicle to the object is output. On the other hand, in the object position calculation device 100 according to the seventh embodiment, the forward direction of the vehicle is used as a reference, and the relative angle of the direction in which the object exists is identified and output.

[0142] For example, if the object is a lane marking representing the boundary between driving lanes on a highway, the relative angle of the direction in which the object exists can be defined as the angle between the direction in which the lane markings around the vehicle extend and the direction in which the vehicle is traveling. Furthermore, if the object is a structure with a specific orientation, the angle between the orientation of the structure and the direction in which the vehicle is traveling can be used as relative position information. Thus, determining a relative angle that is corrected by specifying a relative angle from the relative position information is also extremely important for vehicle operation, as in the first embodiment.

[0143] The configuration diagram of an object position calculation device 100 according to the seventh embodiment differs from that of the first embodiment shown in FIG. 1. The first position information acquisition unit 15 transmits the relative angle (first relative angle) to the object acquired by the first position information acquisition unit 15 in the current cycle to the first difference calculation unit 21 and the object position calculation unit 26. The second position information acquisition unit 45 transmits the relative angle (second relative angle) to the object acquired by the second position information acquisition unit 45 in the current cycle to the first difference calculation unit 21. The other processing blocks are the same as those in the first embodiment, except that the relative position information is replaced with a relative angle.

[0144] <Processing of Object Position Calculation Device> Fig. 19 is a first flowchart showing the processing of the object position calculation device 100 according to embodiment 7. Fig. 20 is a second flowchart showing the processing of the object position calculation device 100. Fig. 20 shows a continuation of the processing of Fig. 19 .

[0145] 19 and 20 are executed by the arithmetic processing unit of the object position calculation device 100. This processing may be executed at predetermined time intervals (for example, every 1 ms). Instead of at predetermined time intervals, this processing may be executed in response to an event such as the first classifier 11 and the second classifier 41 of the object position calculation device 100 identifying an object.

[0146] 19 and 20 are obtained by replacing the relative position information in the processes of FIGS. 4 and 5 according to embodiment 1 with relative distances. The processes from step S301 to step S315 are similar to the processes of step S101 to step S115, and therefore will not be described here.

[0147] According to the object position calculation device 100 of the seventh embodiment, the updated correction value calculated by the correction value update unit 23 is stored in the correction value storage unit 25, and then the object position calculation unit 26 corrects the relative angle to the object using the stored correction value. Furthermore, if the difference value of the first position information is an invalid value, the correction value stored in the correction value storage unit 25 is sent to the object position calculation unit 26 without being changed. With this configuration, even if the second position information acquisition unit 45 fails to acquire the relative angle to the object, the corrected relative angle can be calculated using the correction value calculated in a past cycle.

[0148] 15 described in the sixth embodiment, if the object is a lane marking and the second classifier 41 is a visible light camera, the configuration of the seventh embodiment makes it possible to correct an error in the relative angle to the lane marking with respect to the first position information acquisition unit 15, even if the camera fails to measure the lane marking due to a temporary decrease in the amount of light. In other words, the configuration of the seventh embodiment makes it possible to reduce the frequency of an event in which the estimation accuracy of the relative angle to the object, i.e., the vehicle position, decreases, even if the recognition device is temporarily unable to measure valid relative position information.

[0149] Although various exemplary embodiments and examples are described in this disclosure, the various features, aspects, and functions described in one or more embodiments are not limited to the application of a particular embodiment, but may be applied to the embodiments alone or in various combinations. Therefore, countless variations not illustrated are contemplated within the scope of the technology disclosed herein. For example, this includes cases where at least one component is modified, added, or omitted, or where at least one component is extracted and combined with components of another embodiment.

[0150] REFERENCE SIGNS LIST 11 First classifier, 12 Inertial sensor / speedometer, 13 Satellite positioning device, 14, 33 Map database, 15 First position information acquisition unit, 21 First difference calculation unit, 22 Second difference calculation unit, 23 Correction value update unit, 24 Update necessity determination unit, 25 Correction value storage unit, 26 Object position calculation unit, 31 First classifier accuracy information acquisition unit, 32 Second classifier accuracy information acquisition unit, 34 Correction value prediction unit, 41 Second classifier, 45 Second position information acquisition unit, 50 Vehicle, 100 Object position calculation device

Claims

1. An object position calculation device comprising: a first identifier that identifies a fixed object existing outside a moving body; a first position information acquisition unit that acquires first relative position information between the fixed object identified by the first identifier and the moving body; a second identifier that identifies the fixed object; a second position information acquisition unit that acquires second relative position information between the fixed object identified by the second identifier and the moving body; a first difference calculation unit that calculates a difference value between the second relative position information of the fixed object acquired by the second position information acquisition unit and the first relative position information of the fixed object acquired by the first position information acquisition unit, and outputs the difference value of first position information; a second difference calculation unit that calculates a difference value between the difference value of the first position information output by the first difference calculation unit and a correction value of previously updated position information, and outputs the difference value of second position information; a correction value update unit that updates the correction value of the position information based on the difference value of the second position information output by the second difference calculation unit; and an object position calculation unit that calculates the relative position information of the fixed object by correcting the first relative position information based on the correction value of the position information updated by the correction value update unit.

2. An object position calculation device as described in claim 1, further comprising a second classifier accuracy information acquisition unit that acquires identification accuracy information of the second classifier defined for each road section, wherein the correction value update unit increases an update amount of the correction value of the position information when the accuracy of the identification accuracy information acquired by the second classifier accuracy information acquisition unit is higher than a predetermined accuracy determination value, and decreases an update amount of the correction value when the accuracy of the identification accuracy information acquired by the second classifier accuracy information acquisition unit is equal to or lower than the accuracy determination value.

3. The object position calculation device according to claim 2, wherein the second classifier is an image detector, and the classification accuracy information of the second classifier is set so that the classification accuracy for road sections that are tunnels and tunnel exits is lower than that for other road sections.

4. An object position calculation device according to any one of claims 1 to 3, wherein the first classifier is a classifier capable of discriminating a longer distance than the second classifier.

5. The object position calculation device according to claim 4, wherein the first classifier is a radar device using radio waves, and the second classifier is a classifier using light.

6. An object position calculation device according to any one of claims 1 to 5, wherein the first classifier is a classifier having a higher accuracy of detecting the distance to the fixed object than the second classifier, and the second classifier is a classifier having a higher accuracy of detecting the angle of the fixed object than the first classifier.

7. An object position calculation device as claimed in any one of claims 1 to 6, further comprising a second classifier accuracy information acquisition unit that acquires classification accuracy information of the second classifier, wherein the correction value update unit stops updating the correction value when accuracy of the classification accuracy information acquired by the second classifier accuracy information acquisition unit is lower than a predetermined accuracy determination value.

8. An object position calculation device as claimed in any one of claims 1 to 7, further comprising a correction prediction unit which predicts a future update value of the correction value based on a history of updates of the correction value by the correction value update unit, wherein the correction value update unit updates the correction value using the future update value predicted by the correction prediction unit when updating of the correction value has been stopped beyond a predetermined judgment period.

9. The object position calculation device according to claim 8, further comprising a first classifier accuracy information acquisition unit that acquires classification accuracy information of the first classifier, and a correction prediction unit that predicts a future update value of the correction value based on the classification accuracy information of the first classifier acquired by the first classifier accuracy information acquisition unit and a history of updates to the correction value.

10. An object position calculation device described in any one of claims 1 to 4, wherein the first position information acquisition unit acquires first relative position information of the fixed object based on the current position of the moving body and position information of the fixed object defined by map data.

11. An object position calculation device according to any one of claims 1 to 10, wherein the relative position information includes a distance between the moving body and the fixed object.

12. An object position calculation device according to any one of claims 1 to 10, wherein the relative position information includes a relative angle between the moving body and the fixed object.

Images