Object detection device

Abstract

comprising an object detection device (7) mounted in a vehicle: a first area definition unit (71, S11-S12) configured to define a first object area (Rr) containing a first detection point (Pr), wherein the first detection point (Pr) indicates the position of a first object detected using a vehicle-mounted radio measuring device (2) relative to a reference point (Po) on an XY plane, wherein an X-axis direction of the XY plane is a vehicle width direction and a Y-axis direction of the XY plane is a vehicle length direction; a second area definition unit (72, S13-S14) configured to define a second object area (Ri) containing a second detection point (Pi), wherein the second detection point (Pi) indicates the position of a second object detected on the basis of an image taken by a vehicle-mounted monocular camera (3) relative to the reference point (Po) on the XY plane, wherein the second object area (Ri) is defined by a second azimuth area of ​​an azimuth angle of the second object from the forward direction of the vehicle; wherein the second area definition unit (72, S145-S147) is configured to determine whether the second object is a pedestrian or a vehicle, and if it has been determined that the second object is a pedestrian, to increase the second azimuth area compared to when it has been determined that the second object is a vehicle; and a determination unit (73, S15-S16) that is configured to determine whether or not there is an overlap of the first and second object areas (Rr, Ri) on the XY plane, and if it has been determined that there is an overlap of the first and second object areas (Rr, Ri) on the XY plane, to determine that the first and second objects are the same.

Description

BACKGROUND OF THE INVENTION (Technical Field)

[0001] The present invention relates to the detection of an object using a radio measuring device of a camera. (State of the art)

[0002] DE 60 2004 004 911 T2 discloses an object detection system comprising: a radar detection device which detects an object using a radar, an image detection device which detects an object using an image, and a comparison device which performs a comparison between a detection result of the radar detection device and a detection result of the image detection device to determine whether an identical object is detected by the radar detection device and the image detection device, wherein the comparison device performs a first comparison between an object which is detected in a current comparison by the radar detection device and an object which has been determined to be detected by the radar detection device and the image detection device in a previous comparison;performs a second comparison between an object detected by the image capture device in a current comparison and an object determined in the previous comparison to be detected by the radar capture device and the image capture device if the previous comparison determined that the identical object is detected by the radar capture device and the image capture device; and determines, based on the first and second comparisons, whether the radar capture device and the image capture device detect the identical object.

[0003] DE 10 2006 026 724 A1 discloses an image processing device for a vehicle, including a radar for transmitting a radio wave to the outside of the vehicle, wherein detection of an object in a first area outside the vehicle occurs using a reflected wave of the transmitted radio wave. The image processing device further includes: a camera for capturing an image in a second area including the first area; and an image processing unit for processing the captured image to detect the object detected by the radar in the captured image. A view outside the vehicle is detected based on the detection result of the radar and the detection result of the camera.

[0004] DE 60 2005 004 365 T2 discloses an image processing system for mounting on a vehicle, the image processing system comprising: a radar configured to measure distance and direction to an object outside the vehicle based on reflected electromagnetic waves emitted therefrom by scanning outside the vehicle; an image pickup device that acquires an external image of the vehicle; and an image processor for performing image processing on a specified image processing area in an image acquired with the image pickup device.wherein the image processor serves to determine a center position of the image processing area according to measurement points of an object detected by the radar and the size of the image processing area according to a beam profile of electromagnetic waves output by the radar, characterized in that the image processor serves to, when a number of measurement points are located at the same distance and at locations within intervals in the horizontal direction, the intervals being equal to or smaller than the resolution of the radar, group the number of measurement points as being associated with a single object and to connect image processing areas individually associated with the measurement points as a single image processing area;wherein the electromagnetic waves form a beam having a cross-sectional shape of a circle or an ellipse and propagating at an angle of propagation smaller than the angle of view of the image pickup device in all directions; and that the image processor determines the size of the image processing area according to half-intensity locations where the beam intensity of electromagnetic waves output by the radar is half the maximum intensity of the electromagnetic waves, and decreases the size of the image processing area as the distance from the object detected by the radar increases.

[0005] DE 197 43 255 B4 discloses a moving object detection system for detecting an object moving in the direction of the forward movement of a vehicle, which is equipped with the moving object detection system, in order to prevent a collision between the moving object and the vehicle, wherein said moving object detection system comprises: detection means for detecting a defined forward field of view including a path along which the vehicle is moving, in order to collect data on moving objects in said area in front of the vehicle based on detection echoes, and detection means for determining a lateral speed of a moving object moving transversely to the path of travel of the vehicle and a longitudinal distance of said moving object in said direction of travel, based on said data,to determine a deviation from the stated longitudinal distance.,

[0006] DE 10 2009 033 853 A1 discloses a method for operating a driver assistance system with a plurality of sensors and a device for implementing the method. According to the invention, the complexity of a traffic situation detected by the sensors is determined based on sensor data from the sensors, and the reliability of the sensor data for each sensor is determined depending on the determined complexity of the traffic situation and / or depending on the sensor data from at least one sensor.

[0007] A vehicle collision avoidance system is required to accurately detect an object such as a vehicle other than a controlled vehicle (i.e., a vehicle in which the system is mounted) or a pedestrian. A vehicle collision avoidance system, as disclosed in Japanese Patent Application Publication No. JP 2006-292475 A, is configured to detect an object using a radar and a camera. Specifically, the disclosed vehicle collision avoidance system uses a millimeter-wave radio detector and a stereoscopic camera separately, and determines that an object detected by the millimeter-wave radio detector and an object detected by the stereoscopic camera are the same when a positional relationship therebetween satisfies a predetermined criterion.

[0008] It is also possible for a vehicle collision avoidance system to use a monocular camera instead of the stereoscopic camera as an alternative to the disclosed vehicle collision avoidance system. However, the monocular camera and the stereoscopic camera differ significantly in terms of the principle of detecting a position, in particular a forward distance of an object relative to the controlled vehicle, and the accuracy of detecting the forward distance of the object by the monocular camera is significantly lower than the accuracy of detecting the forward distance of the object by the stereoscopic camera. Therefore, replacing the stereoscopic camera with the monocular camera may not lead to a satisfactory detection result regarding the object.

[0009] In view of the foregoing, it is therefore an object of the present invention to provide an object detecting apparatus capable of detecting an object more satisfactorily using a radio measuring device and a monocular camera. SUMMARY OF THE INVENTION

[0010] This object is achieved by the object detection device having the features of claim 1. Further advantageous developments and embodiments of the invention are the subject of the subsequent claims.

[0011] According to an exemplary embodiment of the present invention, there is provided an object detection device mounted in a vehicle, including a first area defining unit, a second area defining unit, and a determination unit.

[0012] The first area definition unit is configured to define a first object area including a first detection point indicating a position of a first object detected using a vehicle-mounted radio measurement device relative to a reference point on an XY plane. An X-axis direction of the XY plane is a vehicle width direction, and a Y-axis direction of the XY plane is a vehicle length direction.

[0013] The second area definition unit is configured to define a second object area including a second detection point indicating a position of a second object detected based on an image captured by the vehicle-mounted monocular camera with respect to the reference point on the XY plane.

[0014] The determination unit is configured to determine whether or not there is an overlapping area of ​​the first and second object areas on the XY plane, and when it is determined that there is an overlapping area of ​​the first and second object areas on the XY plane, to determine that the first and second objects are the same.

[0015] With this configuration, it is possible to determine that the first and second objects are the same even if the first and second detection points do not completely coincide. This prevents a false determination in which the first and second objects are different objects. Furthermore, it is likely that the first and second objects are incorrectly determined to be the same when the first and second objects are actually different objects.

[0016] In the device according to the present embodiment, the second object region is defined by a second azimuth range of an azimuth angle of the second object from the forward direction of the vehicle. This can prevent erroneous determinations. That is, accuracy in detecting the azimuth angle of the second object based on the captured image from the monocular camera is relatively high. Defining the second object region using such a second azimuth range of the azimuth angle results in the second object region being suitable for the characteristics of the monocular camera, compared to cases where the second object region is defined by an X-coordinate range whose width is constant across X-coordinates. This can prevent erroneous determinations. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings show: Fig. 1A is a schematic block diagram of a collision mitigation device according to a first embodiment of the present invention; Fig. 1B is a schematic block diagram of a collision mitigation ECU of the first embodiment; Fig. 2 is a flowchart showing a processing procedure performed according to the collision mitigation ECU of the first embodiment; Fig. 3 Error ranges defined according to the first embodiment; Fig. 4 is a flowchart of a process flow for defining a radio measuring device error range according to the first embodiment; Fig. 5 is a flowchart of a process flow for defining an image defect region according to the first embodiment; Fig. 6 shows error regions defined according to a second embodiment; Fig. 7 shows error regions defined according to a third embodiment; Fig. 8 shows a radio measuring device error range defined according to the third embodiment; Fig. 9A a relationship between θr and Eθ1; Fig. 9B a relationship between r1 and Eθ2; Fig. 10 shows an image defect region defined according to the third embodiment; Fig. 11 a relationship between r2 and Eθi; Fig. 12 Error ranges defined according to a fourth embodiment; Fig. 13 shows an image defect region defined according to a fifth embodiment; Fig. 14 is a flowchart showing a process flow for defining an image defect region of the fifth embodiment; Fig. 15A-15C a stationary and a moving pedestrian; and Fig. 16 shows an image defect region defined according to a modification of the fifth embodiment. DESCRIPTION OF SPECIFIC EMBODIMENTS

[0018] The present inventions will be described more clearly below with reference to the accompanying drawings. Like elements are designated by like reference numerals throughout. (First embodiment)

[0019] Fig. 1A illustrates a schematic diagram of a collision mitigation device 1 according to a first embodiment of the present invention. The collision mitigation device 1 is mounted in a vehicle (hereinafter referred to as a reference vehicle) and includes a millimeter-wave radio sensing device 2, a monocular camera 3, a brake control unit (ECU) 4, an engine ECU 5, an alarm unit 6, and a collision mitigation ECU 7. In the collision mitigation device 1, the collision mitigation ECU is communicatively connected to, for example, the millimeter-wave radio sensing device 2, the monocular camera 3, the brake ECU 4, the engine ECU 5, and the alarm unit 6.

[0020] The millimeter-wave radio sensing device 2 is mounted in the front center portion of the host vehicle to detect objects such as other vehicles and pedestrians using millimeter waves. The millimeter-wave radio sensing device 2 transmits millimeter waves from the host vehicle forward while scanning the surroundings in the horizontal plane and receiving reflected millimeter waves, thereby transmitting transmitted and received data in the form of radio signals to the collision mitigation ECU 7.

[0021] The monocular camera 3 contains a charge-coupled device (CCD) camera and is mounted on a front center portion of the reference vehicle. The monocular camera 3 transmits data from captured images in the form of image signals to the collision mitigation ECU 7.

[0022] The brake ECU 4 includes a CPU, ROM, RAM, and other brake controllers for the host vehicle. Specifically, the brake ECU 4 controls a brake actuator (brake ACT) in response to a detection value output from a sensor for detecting a brake pedal depression amount. The brake actuator serves as an actuator that opens and closes a pressure-increasing control valve and a pressure-reducing control valve provided in a brake hydraulic circuit. The brake ECU 4 controls the brake actuator according to instructions from the collision mitigation ECU 7 to increase a braking force on the host vehicle.

[0023] The engine ECU 5 includes a CPU, ROM, RAM, and other devices for controlling engine start-up and shutdown, fuel injection amount, ignition timing, and other operations. Specifically, the engine ECU 5 controls a throttle actuator (throttle ACT) in response to a detection value output from a sensor for detecting an accelerator pedal depression amount. The throttle actuator serves as an actuator that opens and closes a throttle valve provided in an air intake passage. The engine ECU 5 controls the throttle actuator according to instructions from the collision mitigation ECU 7 to reduce the driving force of the internal combustion engine.

[0024] The alarm unit 6 alerts a driver of the reference vehicle acoustically and visually after receiving a warning signal from the collision mitigation ECU 7.

[0025] The collision mitigation ECU 7 includes a CPU, ROM, RAM, and other devices to integrally control the collision mitigation device 1. The collision mitigation ECU 7 receives radio signals from the millimeter-wave radio measuring device 2 and image signals from the monocular camera 3 at every predetermined time interval based on a corresponding clock.

[0026] Next, an object detection process performed in the collision mitigation device 1 will be explained. An object detection program, that is, a program for executing the object detection process, is stored in the ROM or the like of the collision mitigation device 1. Fig. 2 illustrates a flowchart of the object detection process performed in the collision mitigation device 1 at every predetermined time interval.

[0027] First, in step S11, the collision mitigation ECU 7 detects an object based on a radio signal transmitted from the millimeter-wave radio measuring device 2 (i.e., detection information from the millimeter-wave radio measuring device 2). More specifically, based on the radio signal, the collision mitigation ECU 7 calculates or determines a linear distance from the host vehicle to the object and a horizontal azimuth angle of the object (i.e., an angular position of the object from the forward direction of the host vehicle). Based on these calculated values, the collision mitigation ECU 7 calculates or determines, as shown in Fig. 3, the position coordinates (X and Y coordinates) of the object on the XY plane are defined as a detection point Pr of the object on the XY plane. The X axis of the XY plane is defined by a vehicle width direction (transverse direction) of the reference vehicle, and the Y axis of the XY plane is defined by a vehicle longitudinal direction (forward direction) of the reference vehicle. A reference point Po of the XY plane is set at a nose (or front end) of the reference vehicle on which the millimeter-wave radio measuring device 2 is mounted. The detection point Pr is a position of the object relative to the reference point Po. It should be noted that Fig. 3 illustrates an example of an object located in front of and on the right side of the vehicle. Furthermore, in step S11, the collision mitigation ECU 7 may calculate a speed of the object relative to the host vehicle in addition to the detection point Pr of the object. Subsequently, the object detected in step S11 is referred to as a "radio object."

[0028] Subsequently, in step S12, the collision mitigation ECU 7 defines an error range Rr centrally located at a detection point Pr detected in step S11, as shown in Fig. 3. More specifically, an X-coordinate range of the error range Rr centrally located at an X-coordinate of the detection point Pr and a Y-coordinate range of the error range Rr centrally located at the X-coordinate of the detection point Pr are assumed to be assumed errors in the X and Y coordinates, respectively, which are determined based on the characteristics of the millimeter-wave radio measuring device 2.

[0029] The error range Rr can be expressed by Xr-EXr≤X≤Xr+EXr and Yr-EYr≤Y≤Yr+EYr, where Xr, Yr are the X and Y coordinates of the detection point Pr, respectively, and ±EXr, ±EYr are assumed errors in the X and Y coordinates, respectively.

[0030] The collision mitigation ECU 7 changes the assumed error EXr depending on a radio object type, such as a vehicle, a pedestrian, or the like. More specifically, Fig. 4 illustrates a process flow performed in step S12. The collision mitigation ECU 7 determines whether the radio object is a vehicle or not in step S121. If it is determined in step S121 that the radio object is not a vehicle (but a pedestrian or the like), the collision mitigation ECU 7 sets the assumed error EXr to a standard value IXr in step S122. If it is determined in step S121 that the radio object is a vehicle, the collision mitigation ECU 7 sets the assumed error EXr to the standard value IXr multiplied by a constant C1 (greater than 1) in step S123.

[0031] When the collision mitigation ECU 7 determines that the radio object is a vehicle, the assumed error EXr is increased (i.e., the error range Rr is expanded in the X direction while keeping the height constant) compared to a case where the collision mitigation ECU 7 determines that the radio object is a pedestrian. This is because a lateral range of points of a vehicle from which millimeter waves can be reflected is larger than a lateral range of points of a pedestrian from which millimeter waves can be reflected.

[0032] Note that in step S121, it is assumed that the radio object is of the radio object type determined in the previous cycle, since at this stage it cannot yet be determined whether the radio object detected in step S11 is actually a vehicle. Hereinafter, the error range Rr defined in step S12 (i.e., the error range Rr defined for the detection point Pr of the radio object based on the characteristics of the millimeter-wave radio measurement device 2) is referred to as the "radio error range Rr."

[0033] Next, in step S13, the collision mitigation ECU 7 detects an object based on an image signal transmitted from the monocular camera 3 (i.e., a captured image from the monocular camera 3). Specifically, the collision mitigation ECU 7 applies image analysis to the captured image represented by the image signal to identify an object. This identification can be implemented through a matching process flow with prestored object models. An object model is prepared for each type of object, such as a vehicle, a pedestrian, or the like, enabling not only determination of the presence of an object but also identification of the object type.The collision mitigation ECU 7 determines a Y coordinate of the object on the XY plane based on a vertical position of the object in the captured image, and a horizontal azimuth angle of the object (an angular position from the forward direction of the reference vehicle) based on a horizontal position of the object in the captured image.

[0034] When the object is detected as being further away from the reference vehicle in the forward direction of the reference vehicle, that is, when the Y coordinate of the object is increased, a lower end of the object tends to be located at a higher position in the captured image. This allows the Y coordinate of the object to be determined based on the height of the object's lower end position in the captured image. However, inaccurate detection of the object's lower end position leads to reduced accuracy in detecting the object's Y coordinate.

[0035] Furthermore, a horizontal displacement of the object from the focus of extension (FOE) of the monocular camera 3 tends to increase with an increasing angular displacement of the object from the forward direction of the reference vehicle (more precisely, a line of X=0). This allows a horizontal azimuth angle of the object to be determined based on a distance from the FOE to a vertical line passing through the center of the object.

[0036] Thus, in step S13, the collision mitigation ECU 7 determines the Y coordinate and the horizontal azimuth angle (angular position) of the object on the XY plane as the detection point Pi of the object on the XY plane, as shown in Fig. 3. The detection point Pi of the object represents a position of the object relative to the reference point Po. Hereinafter, the object detected in step S13 (the object detected based on the captured image by the monocular camera 3) is referred to as an “image object.”

[0037] Subsequently, in step S14, the collision mitigation ECU 7 defines an error range Ri centrally located at the detection point Pi of the image object calculated in step S13, as shown in Fig. 3. More specifically, a Y-coordinate (vertical) range of the error range Ri centralized at the Y-coordinate of the detection point Pi and a horizontal azimuth angle range of the error range Ri (hereinafter referred to as "image azimuth angle range") centralized at the horizontal azimuth angle of the detection point Pi are respectively assumed errors in the Y-coordinate and the horizontal azimuth angle, which are predetermined based on the characteristics of the millimeter-wave radio measuring device 3.

[0038] The error range Ri can be expressed by Yi-EYi≤Y≤Yi+EYi and θi-Eθi≤θ≤θi+Eθi, where Yi and θi represent the Y coordinate and horizontal azimuth angle of the detection point Pr, and ±EYi and ±Eθi are assumed errors in the Y coordinate and horizontal azimuth angle, respectively. That is, the image azimuth angle range of the error range Ri is 2Eθi wide, and the Y coordinate range of the error range Ri is 2EYi wide.

[0039] The collision mitigation ECU 7 changes the assumed error EYi in the Y coordinate and the assumed error Eθi in the horizontal azimuth angle according to a type and a position of the image object. Fig. 5 illustrates a process performed in step S14. The collision mitigation ECU 7 sets the assumed error EYi in the Y coordinate to a standard value IYi multiplied by the Y coordinate Yi of the image object in step S141. That is, the assumed error EYi in the Y coordinate increases with an increase in the Y coordinate of the detection point Pi of the image object. In the present embodiment, the assumed error EYi in the Y coordinate is proportional to the Y coordinate of the detection point Pi of the image object. This is because an error in the lower end position of the image object in the captured image becomes larger as the object is located at a greater distance from the host vehicle in the forward direction of the host vehicle, that is, as the Y coordinate of the object is increased.

[0040] Next, the collision mitigation ECU 7 determines whether the image object is a vehicle or not in step S142. If it is determined in step S142 that the image object is not a vehicle (but, for example, a pedestrian or the like), the collision mitigation ECU 7 leaves the assumed error EYi unchanged in step S143. If it is determined in step S142 that the image object is a vehicle, the collision mitigation ECU 7 multiplies the assumed error EYi by a constant C2 (greater than one) in step S144. That is, if it is determined that the image object is a vehicle, the assumed error EYi is increased compared to when it is determined that the image object is not a vehicle. This is because, if the object is a vehicle, wheels and the like are likely to be obscured by a vehicle body, and the lower end position of the object can therefore be determined less accurately, i.e., the Y-coordinate of the object can be determined less accurately than in a case where the object is not a vehicle.

[0041] Next, the collision mitigation ECU 7 determines whether the image object is a pedestrian or not in step S145. If it is determined in step S145 that the image object is not a pedestrian (but a vehicle, etc.), the collision mitigation ECU 7 sets the assumed error Eθi to a standard value Iθi in step S146. If it is determined in step S145 that the image object is a pedestrian, the collision mitigation ECU 7 multiplies the standard value Eθi by a constant C3 (greater than one) in step S147. That is, if it is determined that the image object is a pedestrian, the assumed error Eθi is increased compared to a case where it is determined that the image object is a vehicle.This results from the fact that, since the pedestrian detected in front of the reference vehicle is likely to be a pedestrian crossing the road, the horizontal position of the pedestrian can be determined less accurately than if the image object was determined to be a vehicle.

[0042] Hereinafter, the error range Ri defined in step S14 (ie, the error range Ri defined for the detection point Pi of the image object based on the characteristics of the monocular camera 3) is referred to as “image error range Ri”.

[0043] Subsequently, in step S15, the collision mitigation ECU 7 determines whether or not there is an overlap region of the radio error region Rr and the image error region Ri.

[0044] If it is determined in step S15 that an overlapping area of ​​the radio error area Rr and the image error area Ri (a shaded area in Fig. 3), the collision mitigation ECU 7 determines that the radio object and the image object are the same in step S16. A position of the object determined to be the same on the XY plane is a position Pf specified by the Y coordinate Yr of the detection point Pr of the radio object and the horizontal azimuth angle θi of the image object.

[0045] Further, in step S17, the collision mitigation ECU 7 calculates a degree of confidence in determining that the radio object and the image object are the same. In the present embodiment, the degree of confidence is defined by an angular difference between the horizontal azimuth angle of the detection point Pr of the radio object and the horizontal azimuth angle of the detection point Pi of the image object. Such a degree of confidence increases with an increase in the angular difference.

[0046] If it is determined in step S15 that there is no overlap area of ​​the radio error area Rr and the image error area Ri, the collision mitigation ECU 7 determines that the radio object and the image object are not the same, that is, that they are different objects.

[0047] Next, the collision mitigation ECU 7 performs collision mitigation control according to the position Pf of the object and the reliability level calculated in step S17. For example, if the host vehicle is likely to collide with the object, the collision mitigation ECU 7 sends a warning signal to an alarm unit 6 to alert the driver. If there is an even higher probability that the host vehicle will collide with the object, the collision mitigation ECU 7 instructs the engine ECU 5 to reduce a driving force of the internal combustion engine and / or instructs the brake ECU 4 to increase a braking force of the host vehicle. Furthermore, the collision mitigation ECU 7 changes control aspects according to a reliability level. For a high reliability level, for example, a control initiation timing is advanced further than a control initiation timing for a low reliability level.

[0048] The first embodiment can offer the following advantages. (1) In the present embodiment, the X-coordinate range of the image defect area Ri is an image azimuth angle range centrally provided at the horizontal azimuth angle θi of the object detected based on the captured image (image object), thereby preventing misdeterminations. This is because the accuracy in detecting the horizontal azimuth angle of the image object based on the captured image by the monocular camera 3 is relatively high compared to the accuracy in detecting the Y-coordinate of a position of the image object. Therefore, not the X-coordinate range but the image azimuth angle range of the image defect area Ri is set to have a constant width. This makes the image defect area Ri suitable for the characteristics of the monocular camera 3, thereby further preventing misdeterminations. (2) In the present embodiment, the X-coordinate range of the radar error range Rr is increased when it is determined that the radio object is a vehicle, in contrast to the case when it is determined that the radio object is a pedestrian (see steps S121-S123 in Fig. 4). This results in the radio error range Rr being suitable for the characteristics of a vehicle with a larger lateral point area from which millimeter waves can be reflected than that of a pedestrian, thus preventing false determinations. (3) In the present embodiment, the Y-coordinate range of the image defect area Ri is enlarged with an increase or enlargement of the Y-coordinate of the detection point Pi of the image object (see step S141 in Fig. 5). This results in the image error range Ri being suitable for the characteristic values ​​such that an error in the lower end position of the image object in the acquired image becomes larger with an increase in the Y coordinate of the object, which can prevent false determinations. (4) In the present embodiment, when it is determined that the image object is a vehicle, the Y coordinate range of the image defect area Ri is enlarged, compared with the case where it is determined that the image object is a pedestrian (see steps S142-S144 in Fig. 5). This means that the image error range Ri is suitable for parameters where a lower end position of a vehicle cannot be accurately detected, thus preventing incorrect determinations. (5) In the present embodiment, when it is determined that the image object is a pedestrian, the image azimuth angle range of the image defect area Ri is increased, compared with the case where it is determined that the image object is a vehicle (see steps S142-S144 in Fig. 5). This means that the image error range Ri can be suitable for the characteristics of a laterally moving pedestrian (i.e., moving in the X direction), thus preventing false determinations. (6) In the present embodiment, when it is determined that the radio object and the image object are the same, a position of the object on the XY plane is defined by the Y coordinate of the detection point Pr of the radio object and the horizontal azimuth angle θi of the image object. This allows a position of the object to be determined using characteristics from the millimeter-wave radio measurement device 2 and the monocular camera 3. (7) In the present embodiment, when it is determined that the radio object and the image object are the same, a reliability degree is calculated upon determining that the radio object and the image object are the same. This allows vehicle control aspects to be changed according to the calculated reliability degree even when it is determined that the radio object and the image object are the same. More specifically, the reliability degree is defined by an angular difference between the horizontal azimuth angle of the detection point Pr of the radio object and the horizontal azimuth angle of the detection point Pi of the image object. Such a reliability degree becomes larger with an increase in the angular difference. This definition can simplify the calculation of the reliability degree.

[0049] With reference to Fig. 1B, the collision mitigation ECU 7, which serves as an object detection device, includes a first area definition unit 71 responsible for executing steps S11-S12, a second area definition unit 72 responsible for executing steps S13-S14, and a determination unit 73 responsible for executing steps S15-S16. The collision mitigation ECU 7 further includes a reliability degree calculation unit 74 responsible for executing step S17. These units 71-74 may be implemented by the collision mitigation ECU 7 executing appropriate computer programs stored in ROM or the like to function according to the processing procedures explained in connection with the embodiments of the invention.

[0050] The radio object corresponds to a first object. The image object corresponds to a second object. The detection point Pr corresponds to a first detection point. The detection point Pi corresponds to a second detection point. The radio error range Rr corresponds to a first object range. The image error range Ri corresponds to a second object range. The image azimuth angle range corresponds to a second azimuth range. (Second embodiment)

[0051] Next, a second embodiment of the present invention will be explained. In the second embodiment, only the differences from the first embodiment will be explained.

[0052] In the first embodiment described above, the radio error range Rr is defined by an X-coordinate range of an assumed error in the X-coordinate centrally present at the X-coordinate of the detection point Pr and a Y-coordinate range of an assumed error in the Y-coordinate centrally present at the Y-coordinate of the detection point Pr, wherein the assumed errors in the X- and Y-coordinates are predetermined based on the characteristics of the millimeter-wave radio measuring device 2.

[0053] Alternatively, in the present embodiment, a radio error range Rr is defined by a horizontal azimuth angle range of an assumed error in the horizontal azimuth angle central to the horizontal azimuth angle of the detection point Pr and a Y-coordinate range of the assumed error in the Y-coordinate central to the Y-coordinate of the detection point Pr, wherein the assumed errors in the horizontal azimuth angle and the Y-coordinate are predetermined based on the characteristics of the millimeter-wave radio measuring device 2.

[0054] The radio error range Rr can be expressed by Yr-EYr≤Y≤Yr+EYr and θr-Eθr≤θ≤θr+Eθr, where Yr and θr represent the Y coordinate and horizontal azimuth angle of the detection point Pr, and ±EYr and ±θr are the assumed errors in the Y coordinate and horizontal azimuth angle. That is, the horizontal azimuth angle range of the error range Rr central to the horizontal azimuth angle of the detection point Pr is 2Eθr wide (hereinafter referred to as the "radio azimuth angle range"), and the Y coordinate range of the radio error range Rr central to the Y coordinate of the detection point Pr is 2EYr wide.

[0055] With this configuration, similar advantages to those of the first embodiment can be achieved in the present embodiment. Furthermore, in the present embodiment, the radio error range Rr is defined based on more characteristics of the millimeter-wave radio measurement device 2, thereby preventing incorrect determinations. The radio azimuth angle range corresponds to a first azimuth range. (Third embodiment)

[0056] Next, a third embodiment of the present invention will be explained. With reference to the third embodiment, only differences from the second embodiment will be explained.

[0057] In the second embodiment, as described above, the radio error range Rr and the image error range Ri are defined in a similar manner. As in Fig. As shown in Figure 6, the radio error range Rr is defined by a horizontal azimuth angle range of an assumed error in the horizontal azimuth angle centrally present at the horizontal azimuth angle of the detection point Pr and a Y-coordinate range of an assumed error in the Y-coordinate centrally present at the Y-coordinate of the detection point Pr. Similarly, the image error range Ri is defined by a horizontal azimuth angle range of an assumed error in the horizontal azimuth angle centrally present at the horizontal azimuth angle of the detection point Pi and a Y-coordinate range of an assumed error in the Y-coordinate centrally present at the Y-coordinate of the detection point Pi.

[0058] Alternatively, in the third embodiment, the radio error region Rr and the image error region Ri are defined similarly as follows. The position of the detection point Pr is specified by a linear distance r1 from the reference point Po to the detection point Pr (also referred to as the radial coordinate) and the horizontal azimuth angle of the detection point Pr, as defined in the first and second embodiments.

[0059] In the present embodiment, as shown in Fig. As shown in Figure 8, the radio error range Rr is defined by a horizontal azimuth angle range of the assumed error in the horizontal azimuth angle centrally present at the horizontal azimuth angle of the detection point Pr, and a linear distance range of an assumed error in the linear distance centrally present at the linear distance of the detection point Pr. The assumed errors in the horizontal azimuth angle and the linear distance are predetermined based on the characteristics of the millimeter-wave radio measuring device 2. The radio error range Rr can be expressed by r1-Er1≤r≤r1+Er1 and θr-Eθr≤θ≤θr+Eθr, where r1, θr represent the linear distance and the horizontal azimuth angle of the detection point Pr, and ±Er1, ±θr are assumed errors in the linear distance and the horizontal azimuth angle.That is, the radio error range Rr is defined by the horizontal azimuth angle range from θr-Eθr to θr+Eθr, which is central to the horizontal azimuth angle of the detection point Pr (radio azimuth angle range), and the linear distance range from r1-Er1 to r1+Er1, which is central to the linear distance of the detection point Pr (hereinafter referred to as “radio distance range”).

[0060] The assumed error Eθr in the horizontal azimuth angle θr of the detection point Pr of the radio object is calculated as the product of Eθ1 and Eθ2 (ie, Eθr=Eθ1×Eθ2). As in Fig. 9A, Eθ1 is set to a lower limit θc1 [degrees] for the horizontal azimuth angle θr less than or equal to θr1 [degrees], increased with an increasing horizontal azimuth angle θr from θr1 [degrees] to θr2 [degrees] (θr2>θr1), where Eθ1 - θc1 is proportional to θr - θr1, and set to an upper limit θc2 [degrees] for the horizontal azimuth angle θr greater than or equal to θr2 [degrees].

[0061] With the assumed error Eθr and the use of Eθ1, when the horizontal azimuth angle θr is greater than θr1 [degrees], the assumed error Eθr is increased compared to when the horizontal azimuth angle θr is equal to or less than θr1 [degrees]. As a result, the assumed error Eθr reflects the characteristics of the millimeter-wave radio measuring device 2 in such a way that a detection error is increased with an increased horizontal azimuth angle θr, that is, the antenna gain decreases.

[0062] In addition, as in Fig. As shown in Figure 9B, Eθ2 is set to be indirectly proportional to the linear distance r1 from the reference point Po to the radio target, i.e., Eθ2=rc / r1, where rc is a constant. For linear distance r1 greater than or less than rc [m], Eθ2 is set to the lower limit value θc1 [degrees].

[0063] With this definition of the assumed error Eθr, calculated by multiplying Eθ2 by Eθ1, when the linear distance r1 is smaller than the predetermined value rc [m], the assumed error Eθr in the horizontal azimuth angle is increased compared to the case where the linear distance r1 is greater than or equal to the predetermined value rc [m]. If Eθr were constant over the linear distance from the reference point Po to the detection point Pr, the radio error range Rr would be so narrow for short linear distances (especially in the X-axis direction) that an overlap region between the radio error range Rr and the image error range Ri is unlikely, which would lead to false determinations that the radio object and the image object are different objects, although they are actually the same.Therefore, in the present embodiment, the assumed error Eθr increases as the linear distance from the reference point Po to the detection point Pr decreases, which can prevent the assumed error Eθr from becoming too small. In the present embodiment, the upper limit of the assumed error Eθr is set to θc3 [degrees] (θc3>θc2), and the lower limit of the assumed error Eθr is set to θc1 [degrees].

[0064] The assumed linear distance error Er1 is calculated, for example, as Er1=r1×A1+B, where A1 and B are constants. This means that the assumed error Er1 increases with an increase in the linear distance r1 from the reference point Po to the detection point Pr of the radio target. As a result, the assumed error Er1 reflects the characteristics of the millimeter-wave radio measurement device 2 in such a way that the linear distance detection error increases with an increase in the linear distance r1.

[0065] In addition, as in Fig. 10, the image error range Ri is defined by a horizontal azimuth angle range of an assumed error in the horizontal azimuth angle centrally present at the horizontal azimuth angle θi of the detection point Pi of the image object, and a linear distance range of an assumed error in the linear distance centrally present at the linear distance r2 of the detection point Pi, wherein the assumed errors in the horizontal azimuth angle and the linear distance are predetermined based on the characteristics of the monocular camera 3. The image error range Ri can be expressed by r2-Er2≤r≤r2+Er2 and θi-Eθi≤θ≤θi+Eθi, where r2, θi represent the linear distance and the horizontal azimuth angle of the detection point Pi, and ±Er2, ±θi are the assumed errors in the linear distance and the horizontal azimuth angle.That is, the image error range Ri is defined by the horizontal azimuth angle range from θi-Eθi to θi+Eθi, which is centralized at the horizontal azimuth angle θi of the detection point Pi (image azimuth angle range), and the linear distance range from r2-Er2 to r2+Er2, which is centralized at the linear distance r2 of the detection point Pi (hereinafter referred to as “image distance range”).

[0066] The assumed error Eθi in the horizontal azimuth angle θi of the detection point Pi of the image object is defined in a similar way to the assumed error Eθ2. As in Fig. As shown in Figure 11, Eθi is set to be indirectly proportional to the linear distance r2 from the reference point Po to the detection point Pi of the image object, that is, Eθi = rc / r2. For the linear distance r2 equal to or greater than rc [n], Eθi is set to the lower limit θc1 [degree].

[0067] With this definition, when the linear distance r2 from the reference point Po to the detection point Pi of the image object is smaller than the predetermined value c [n], the assumed error Eθi in the horizontal azimuth angle is increased compared to the case where the linear distance r2 is equal to or larger than the predetermined value rC [m]. In the present embodiment, the upper limit Eθi is set to θc3 [degrees]. When defined without multiplication by Wθ1, the image azimuth range is reduced in width compared to the radio azimuth range. Regarding a detection accuracy of the horizontal azimuth angle θi based on the captured image, it is unlikely to be affected by the horizontal azimuth angle θi itself compared to the horizontal azimuth angle θr.

[0068] The assumed error Er2 in the linear distance r2 is calculated, for example, by Er2 = r3 x A2 + B, where A2 (greater than A1) and B are constants. That is, the assumed error Er2 increases with the increasing linear distance r2 from the reference point Po to the detection point Pi of the image radio object. The magnification rate A2 of the assumed error Gr2 is greater than the magnification rate A1 of the assumed error Er1. This results in the assumed error Er2 representing the characteristics of the monocular camera 3 such that the detection error in the linear distance r2 determined using the monocular camera 3 is greater than the detection error in the linear distance r1 determined by the millimeter-wave radio measuring device 2, and the detection error in the linear distance R2 increases with the increase of the linear distance r2.

[0069] With this configuration, the present embodiment can achieve similar advantages to the second embodiment. The present embodiment can also achieve the following advantages. (1) In the present embodiment, the image defect range Ri is defined by the image azimuth range of a presumed defect and the image distance range of a presumed defect. As a result, the image defect range Ri reflects detection errors more appropriately than an image defect range defined using the Y-coordinate range of a presumed defect instead of the image distance range of a presumed defect. (2) In the present embodiment, the image distance range of an assumed error is increased with an increase in the linear distance r2 from the reference point Po to the detection point of the image object. This results in the image error range Ri being more suitable for the characteristics of the monocular camera 3. (3) In the present embodiment, when the linear distance r2 from the reference point Po to the detection point of the image object is smaller than the predetermined value Rc, the image azimuth range of the assumed error is increased compared to the case where the linear distance r2 is greater than or equal to the predetermined value rc. This can prevent erroneous determinations regarding the presence of objects near the reference vehicle. (4) In the present embodiment, the radio error range Rr is defined by the radio azimuth range of a presumed fault and the radio distance range of a presumed fault. This results in the radio error range Rr reflecting detection errors more appropriately than a radio error range defined using the X-coordinate range of a presumed fault instead of the radio distance range of a presumed fault. (5) In the present embodiment, the radio distance range of the assumed error is increased as the linear distance r1 from the reference point Po to the detection point of the radio object increases. This results in the radio error range Rr being more suitable for the characteristics of the millimeter-wave radio measurement device. (6) In the present embodiment, when the linear distance r1 from the reference point Po to the detection point of the radio object is smaller than the predetermined value Rc, the radio azimuth range of the assumed error is increased compared to the case where the linear distance r1 is greater than or equal to the predetermined value. This results in the radio error range Rr being more suitable for the characteristics of the millimeter-wave radio measurement device 2. (7) In the present embodiment, when the horizontal azimuth angle θr is larger than the predetermined value θr1, the radio azimuth range of an assumed error is increased compared to the case where the horizontal azimuth angle θr is less than or equal to the predetermined value θr1. This results in the radio error range Rr being more suitable for the characteristics of the millimeter-wave radio measurement device 2. (8) In the present embodiment, the radio distance range of the assumed defect is reduced compared to the image distance range of an assumed defect. As a result, the radio defect range Rr and the image defect range Ri more appropriately reflect differences in characteristics between the millimeter-wave radio measurement device 2 and the monocular camera 3. (9) In the present embodiment, the image azimuth range of a presumed fault is restricted compared to the radio azimuth range of a presumed fault. As a result, the radio fault range Rr and the image fault range Re more appropriately reflect differences in characteristics between the millimeter-wave radio measurement device 2 and the monocular camera 3.

[0070] In the present embodiment, as in the first or second embodiment, a position of the object determined to be the same on the XY plane may be a position Pf determined by the X coordinate Xr of the detection point Pr of the radio object and the horizontal azimuth angle θi of the image object. Alternatively, the linear distance r1 may be used instead of the Y coordinate Yr. That is, a position of the object determined to be the same on the XY plane may be set to a position determined by the radial coordinate of the detection point Pr of the radio object and the horizontal azimuth angle θi of the image object.

[0071] The radio distance range corresponds to a first linear distance range. The image distance range corresponds to a wide linear distance range. (Fourth embodiment)

[0072] Next, a fourth embodiment of the present invention will be explained. With reference to the fourth embodiment, only differences from the first embodiment will be explained.

[0073] In the first embodiment described above, the radio error region Rr is defined with reference to the detection point Pr of the radio object. If it is determined that there is an overlap region of the radio error region Rr and the image error region Ri, it is determined that the radio object and the image object are the same.

[0074] Alternatively, in the present embodiment, as shown in Fig. 12, if it is determined that the detection point Pr of the radio object overlaps with the image defect region Ri, that is, if it is determined that the detection point Pr of the radio object exists in the image defect region Ri, it is determined that the radio object and the image object are the same. The present embodiment can be understood as a limiting example of the first embodiment in which the radio defect region Rr is shrunk to the detection point Pr.

[0075] The present embodiment can achieve similar advantages to the first embodiment. The present embodiment achieves the additional advantage of eliminating the process in step S12.

[0076] The detection point Pr corresponds not only to the first detection point but also to the first object area. The present embodiment is applied to the first embodiment. Alternatively, the present embodiment can also be applied to the second or third embodiment. (Fifth embodiment)

[0077] Next, a fifth embodiment of the present invention will be explained. With reference to the fifth embodiment, only differences from the first embodiment will be explained.

[0078] In the first embodiment described above, if it is determined in step S145 that the image object is a pedestrian, the assumed error Eθi in the horizontal azimuth angle is set to a standard value Iθi multiplied by a constant C3 in step S147. Specifically, the horizontal azimuth angle range in the first embodiment is from θi-Eθi to θi+Eθi, and is centralized at the horizontal azimuth angle θi of the detection point Pi (that is, symmetrical about the horizontal azimuth angle θi).

[0079] Alternatively, in the present embodiment, as shown in Fig. 13 illustrates that when it is determined that the image object is a pedestrian, the horizontal azimuth angle range is asymmetrical around the horizontal azimuth angle θi, that is, it contains left and right assumed errors EθiL, EθiR in the horizontal azimuth angle with respect to the horizontal azimuth angle θi according to a pedestrian's movement direction, that is, depending on a left or right movement direction. The left and right assumed errors EθiL, EθiR may differ.

[0080] More specifically, in the fifth embodiment, a process flow that is Fig. 14, as an alternative to the process flow shown in Fig. 5. Since the operations in steps S241-S246 and S248 are similar to the operations in steps S141-S147, corresponding explanations for the operations in steps S241-S246 and S248 are not repeated.

[0081] If it is determined in step S245 that the image object is a pedestrian, the collision mitigation ECU 7 determines whether the pedestrian is moving or not in step S247. If it is determined in step S247 that the pedestrian is not moving, that is, the pedestrian is standing still, the collision mitigation ECU 7 sets the assumed error Eθi in the horizontal azimuth angle to a standard value Iθi multiplied by a constant C3 (greater than 1) in step S248 and step S147. If it is determined in step S247 that the pedestrian is moving, the collision mitigation ECU 7 determines in step S249 whether the pedestrian is moving to the left or not (a first direction).If it is determined in step S249 that the pedestrian is moving to the left, the collision mitigation ECU 7 sets a left assumed error EθiL in the horizontal azimuth angle to the standard value Iθi multiplied by a constant C4 in step S250, and further sets a right assumed error EθiR in the horizontal azimuth angle to the standard value Iθi multiplied by a constant C5 in step S250. The left assumed error EθiL is an assumed error in the horizontal azimuth angle on the left side of the horizontal azimuth angle θi of the detection point Pi of the image object, and the right assumed error EθiR is an assumed error in the horizontal azimuth angle on the right side of the horizontal azimuth angle θi of the detection point Pi of the image object. As shown in FIG. Fig. As shown in Figure 13, the image defect range Ri is not defined by θi-Eθi≤θ≤θi+Eθi, but by θi-EθiL≤θ≤θi+EθiR. The constant C5 is set larger than the constant C4, that is, C5 > C4.

[0082] If it is determined in step S249 that the pedestrian is not moving to the left but to the right (a second direction), the collision mitigation ECU 7 sets the left assumed error EθiL in the horizontal azimuth angle to the standard value Iθi multiplied by a constant C5 in step S251, and further sets the right assumed error EθiR in the horizontal azimuth angle to the standard value Iθi multiplied by a constant C4 in step S251.

[0083] That is, if it is determined that the image object is a pedestrian moving to the left, the right assumed error is set larger than the left assumed error. If it is determined that the image object is a pedestrian moving to the right (i.e., in the X-axis direction), the left assumed error is set larger than the right assumed error.

[0084] The present embodiment configured as above can achieve similar advantages to the first embodiment. The present embodiment can achieve the following additional advantages. The present embodiment makes the image error range Ri suitable for the characteristics of a lateral movement of a pedestrian (ie, a movement in the X direction), thereby preventing erroneous determinations. As described above, the horizontal azimuth angle θi of the image object is determined based on a distance from the FOE to a vertical line passing through the center of the image object. However, for the lateral movement of the pedestrian, as shown in the Fig. 15B, Fig. 15C, the vertical line is more likely to deviate from the actual center of the object, as compared to the vertical line passing through the center of the stationary object, as in Fig. 15A. Since the deviation of the vertical line from the actual center of the object to the left or right of the center of the image object depends on the direction of movement of the pedestrian, the left and right assumed errors are different depending on the direction of movement of the pedestrian.

[0085] In the present embodiment, the assumed error in the direction opposite to the pedestrian's direction of movement is set to be greater than the assumed error in the direction of movement. In some alternative embodiments that differ from the present embodiment, the assumed error in the pedestrian's direction of movement may be set to be greater than the assumed error in the direction opposite to the pedestrian's direction of movement, as shown in Fig. 16. Furthermore, in the present embodiment, the left and right assumed errors differ from each other depending on the pedestrian's moving direction. Alternatively, the left and right assumed errors may be differentiated not only depending on the pedestrian's moving direction, but also depending on the pedestrian's moving speed. For example, when the pedestrian moves in the left direction, the right assumed error increases more as the pedestrian's moving speed increases, compared to the left assumed error. (Other embodiments)

[0086] Next, further embodiments of the first to fifth embodiments will be explained, which can be carried out without departing from the spirit and scope of the present invention. (1) In each of the first to fifth embodiments, when it is determined that there is an overlap region of the radio error region Rr and the image error region Ri, it is determined that the radio object and the image object are the same. Alternatively, when it is determined that the radio error region Rr is included in the image error region Ri, or when it is determined that the image error region Ri is included in the radio error region Rr, it may be determined that the radio object and the image object are the same. (2) In each of the first to fifth embodiments, the degree of reliability in determining that the radio object and the image object are the same is defined by an angular difference between the horizontal azimuth angle of the detection point Pr of the radio object and the horizontal azimuth angle of the detection point Pi of the image object. Alternatively, for example, the degree of reliability in determining that the radio object and the image object are the same may be defined such that it becomes larger when an area of ​​an overlapping area of ​​the radio error area Rr and the image error area Ri (for example, an area of ​​the shaded area in Fig. 3) is increased. Such a degree of reliability can prevent false determinations that the radio object and the image object are the same, even though the horizontal azimuth angles of the radio object and the image object are close to each other, but their linear distances are significantly different. (3) In each of the first to fifth embodiments, the millimeter-wave radio measurement device 2 is used as the radio measurement device. Alternatively, another type of radio measurement device or radar may be used as the radio measurement device. (4) In each of the first to fifth embodiments, the present invention is applied to a collision mitigation device 1. Alternatively, the present invention may be applied to another type of vehicle control. (5) In each of the first to fifth embodiments, the collision mitigation ECU 7 as a controller is a separate element from the millimeter-wave radio measuring device 2 and the monocular camera 3. Alternatively, the collision mitigation ECU 7 may be included in the millimeter-wave radio measuring device 2 and / or the monocular camera 3. (6) In each of the first to fifth embodiments, the pedestrian and the vehicle are detected as exemplary objects. Alternatively, other objects may also be detected. (7) Each component of the present invention represents a certain concept. Therefore, for example, functions of one of the components in each of the first to fifth embodiments may be distributed across a plurality of components, or functions of some components in each of the first to fifth embodiments may be integrated or combined with another component. Furthermore, some of the features of one of the first to fifth embodiments may be merged with or replaced by features of another of the first to fifth embodiments.

[0087] One skilled in the art will understand that the above-described embodiments may be implemented by hardware, software, or a combination of software and hardware. The modules and functions described in conjunction with the embodiments of the invention may be implemented in whole or in part by microprocessors or computers suitably programmed to function according to the methods explained in connection with the embodiments of the invention. A device implementing an embodiment of the invention may, for example, be a computer suitably programmed to execute the above-described mechanisms of the embodiments of the invention.

Claims

[1] Object detection device (7) mounted in a vehicle, comprising: a first area definition unit (71, S11-S12) configured to define a first object area (Rr) including a first detection point (Pr), the first detection point (Pr) indicating a position of a first object detected using a vehicle-mounted radio measuring device (2) relative to a reference point (Po) on an XY plane, wherein an X-axis direction of the XY plane is a vehicle width direction and a Y-axis direction of the XY plane is a vehicle length direction; a second area definition unit (72, S13-S14) configured to define a second object area (Ri) including a second detection point (Pi), the second detection point (Pi) indicating a position of a second object detected based on an image captured by a vehicle-mounted monocular camera (3) relative to the reference point (Po) on the XY plane, the second object area (Ri) being defined by a second azimuth range of an azimuth angle of the second object from the forward direction of the vehicle; wherein the second area definition unit (72, S145-S147) is configured to determine whether the second object is a pedestrian or a vehicle, and when it is determined that the second object is a pedestrian, to enlarge the second azimuth area compared to when it is determined that the second object is a vehicle; and a determination unit (73, S15-S16) configured to determine whether or not there is an overlapping area of ​​the first and second object areas (Rr, Ri) on the XY plane, and when it is determined that there is an overlapping area of ​​the first and second object areas (Rr, Ri) on the XY plane, to determine that the first and second objects are the same. [2] The object detection device (7) according to claim 1, wherein the second area definition unit (72, S249-S251) is configured to determine whether the pedestrian is moving in a first direction along the X-axis or in a second direction opposite to the first direction, and when it is determined that the pedestrian is moving in the first direction, to distinguish a first sub-area of ​​the second azimuth range on the first direction side of the second detection point (Pi) and a second sub-area of ​​the second azimuth range on the second direction side of the second detection point (Pi). [3] Object detection device (7) according to one of claims 1 or 2, wherein the second object range (Ri) is defined by the second azimuth range and a specific Y-coordinate range including a Y-coordinate of the second detection point (Pi) on the XY plane. [4] The object detection device (7) according to claim 3, wherein the second area definition unit (72, S141) is configured to enlarge the Y-coordinate area with an increase in the Y-coordinate of the second detection point (Pi) on the XY plane. [5] The object detection device (7) according to claim 3 or 4, wherein the second area definition unit (72, S142-S144) is configured to determine whether the second object is a pedestrian or a vehicle, and when it is determined that the second object is a vehicle, to enlarge the Y coordinate area compared to when it is determined that the second object is a pedestrian. [6] Object detection device (7) according to one of claims 1 or 2, wherein the second object range (Ri) is defined by the second azimuth range and a second linear distance range, the second linear distance range extending from first to second linear distances from the reference point (Po) on the XY plane and including a linear distance from the reference point (Po) to the second detection point. [7] The object detection device (7) according to claim 6, wherein the second range definition unit (72) is configured to enlarge the second linear distance range with an increase in a linear distance from the reference point (Po) to the second detection point. [8] The object detection device (7) according to claim 6 or 7, wherein the second range definition unit (72) is configured to enlarge the second azimuth range when the linear distance from the reference point (Po) to the second detection point (Pi) is less than a predetermined value, compared to when the linear distance from the reference point to the second detection point (Pi) is greater than or equal to the predetermined value. [9] An object detection device (7) according to any one of claims 1 to 8, wherein the first object range (Rr) is defined by a first azimuth range of an azimuth angle including an azimuth angle of the first object from the forward direction of the vehicle. [10] Object detection device (7) according to claim 9, wherein the first object range (Rr) is defined by the first azimuth range and a first linear distance range, the first linear distance range extending from third to fourth linear distances from the reference point (Po) on the XY plane and including a linear distance from the reference point (Po) to the first detection point (Pr). [11] The object detection device (7) according to claim 10, wherein the first range definition unit (71, S11-S12) is configured to enlarge the first linear distance range with an increase in a linear distance from the reference point (Po) to the first detection point (Pr). [12] The object detection device (7) according to claim 10 or 11, wherein the first range definition unit (71) is configured to enlarge the first azimuth range when the linear distance from the reference point (Po) to the first detection point (Pr) is less than a predetermined value, compared to when the linear distance from the reference point to the first detection point (Pr) is greater than or equal to the predetermined value. [13] The object detection device (7) according to any one of claims 9 to 12, wherein the first area definition unit (71) is configured to enlarge the first azimuth range when the azimuth angle of the first object relative to the reference point (Po) is greater than or equal to a predetermined value, compared to when the azimuth angle of the first object relative to the reference point (Po) is smaller than the predetermined value. [14] The object detection device (7) according to any one of claims 9 to 13, wherein the first area definition unit (71, S121-S123) is configured to determine whether the first object is a pedestrian or a vehicle, and when it is determined that the first object is a vehicle, to enlarge an X-coordinate range of the first object area (Rr) compared to when it is determined that the first object is a pedestrian. [15] The object detection device (7) according to any one of claims 1 to 14, further comprising a reliability degree calculation unit (74, S17) configured to calculate, when it is determined by the determination unit (73) that the first and second objects are the same, a reliability degree in the determination that the first and second objects are the same. [16] An object detection device (7) according to claim 15, wherein the degree of reliability is increased with a reduction in an angular difference between an azimuth angle of the first detection point from the forward direction of the vehicle and an azimuth angle of the second detection point (Pi) from the forward direction of the vehicle. [17] An object detection device (7) according to claim 15 or 16, wherein the degree of reliability is increased with an increase in an area of ​​the overlapping region of the first and second object regions (Rr, Ri). [18] The object detection device (7) according to any one of claims 1 to 17, wherein the determination unit (73) is configured, when it is determined that the first and second objects are the same, to define a position of the object determined to be the same on the XY plane by a Y coordinate of the first detection point (Pr) and an azimuth angle of the second detection point (Pi) from the forward direction of the vehicle. [19] Object detection device (7) according to claim 1, wherein the first object area (Rr) is defined as the first detection point (Pr).

Images