OBJECT DETECTION DEVICE, OBJECT DETECTION METHOD, AND OBJECT DETECTION PROGRAM

Abstract

To provide an object detection device capable of suppressing output of an erroneous fusion object due to erroneous detection by a radar.SOLUTION: An objection detection device comprises: a radar erroneous detection determination unit 4 that sets, as true, a radar erroneous detection suspicion flag for radar-detected object information that is determined to be suspected as erroneous detection; a fusion processing unit 7 that outputs fusion object information based on second sensor-detected object information and the radar-detected object information obtained from the radar erroneous detection determination unit 4; and a radar erroneous detection deletion unit 8 that deletes fusion object information that is suspected as erroneous detection. When a fusion object is determined to be an object detected by only a radar, the fusion processing unit 7 sets, as true, a single-sensor detection flag for the fusion object information, and the radar erroneous detection deletion unit 8 deletes the fusion object information for which the radar erroneous detection suspicion flag is true and the single-sensor detection flag is true.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present application relates to an object detection device, an object detection method, and an object detection program. [Background technology]

[0002] Object detection devices are known that use sensors mounted on vehicles to detect objects around the vehicle. Information about the objects detected by the object detection devices is used to notify or warn the driver, and is also used in autonomous driving systems that control the vehicle to avoid collisions with objects.

[0003] Furthermore, sensor fusion technology is known, which combines information from multiple sensors to fuse the unique information from each sensor and detect objects with higher accuracy. For example, when a millimeter-wave radar (hereinafter simply referred to as "radar") and a camera are installed in front of a vehicle, they are installed so that part of the radar's detection range overlaps part of the camera's detection range. Radar detection devices that detect objects based on radar output can detect objects over long distances and generally have high vertical (depth) ranging accuracy. On the other hand, camera detection devices that detect objects based on camera output can detect obstacles by specializing in specific types through prior learning and generally have high horizontal ranging accuracy. Camera detection devices can also identify object types and output information on the width and length of the object. Although some radar detection devices can output information on object type and size, camera detection devices generally provide higher detection accuracy. When the same object is detected by a radar detection device and a camera detection device, for example, the vertical distance measurement value is highly trusted based on the information output from the radar detection device, and the horizontal distance measurement value and type are highly trusted based on the information output from the camera detection device, and various parameters such as the object's position and speed are calculated and output to a subsequent processing block (control system or alarm system) as fusion object information.

[0004] When multiple sensors are used and information from each sensor is fused to create information on a fusion object, if a false detection occurs in one sensor, the information from that sensor and information from another sensor may be mistakenly determined to be the same object, resulting in output of incorrect information on the fusion object. In this case, an object that does not actually exist may be detected, resulting in, for example, incorrect braking control being performed for a non-existent object. In sensor fusion technology using radar and a camera, an object detection device has been proposed that adds a short-distance flag to the output from the radar and changes the output from the camera depending on the state of the flag to prevent false detections from information from the camera (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 4311861 Summary of the Invention [Problem to be solved by the invention]

[0006] The object detection device shown in Patent Document 1 can suppress the output of erroneous fusion objects due to erroneous detection based on camera information, but has the problem of being unable to suppress the output of erroneous fusion objects due to erroneous detection based on radar information.

[0007] The present application has been made to solve the above-mentioned problems, and aims to provide an object detection device, an object detection method, and an object detection program that suppress the output of erroneous fusion objects due to erroneous detection based on radar information. [Means for solving the problem]

[0008] The object detection device disclosed in the present application detects an object based on information from a first detection device that outputs radar-detected object information including a radar detection position, which is position information of the radar-detected object, for a radar-detected object that is an object detected based on the output of a radar, and information from a second detection device that outputs second-sensor-detected object information including a second-sensor-detected position, which is position information of the second-sensor-detected object, for an object that is detected based on the output of a second sensor different from the radar, and the object detection device includes a radar erroneous detection determination unit that determines whether the radar-detected object information obtained from the first detection device is suspected of being an erroneous detection and sets a radar erroneous detection suspicion flag to true for radar-detected object information that is judged to be suspected of being an erroneous detection, and a fusion object detection unit that determines whether the radar-detected object information obtained from the second sensor is suspected of being an erroneous detection, which includes a fusion object position, which is position information of the fusion object, based on the second-sensor-detected object information and the radar erroneous detection determination unit. and a radar false detection deletion unit that deletes fusion object information that is determined to be suspected of being a false detection. When the fusion processing unit determines that the radar-detected object and the second sensor-detected object are one fusion object, it combines the radar-detected object information and the second sensor-detected object information into one fusion object information and sets the single sensor detection flag of the combined fusion object information to false. When it determines that the fusion object was detected only by radar, it treats the radar-detected object information of the radar-detected object as fusion object information and sets the single sensor detection flag of the fusion object information to true. The radar false detection deletion unit deletes fusion object information for which the radar false detection suspicion flag is true and the single sensor detection flag is true. [Effects of the Invention]

[0009] The object detection device disclosed in the present application detects an object based on information from a first detection device that outputs radar-detected object information including a radar detection position, which is position information of the radar-detected object, for a radar-detected object that is an object detected based on the output of a radar, and information from a second detection device that outputs second-sensor-detected object information including a second-sensor-detected position, which is position information of the second-sensor-detected object, for an object that is detected based on the output of a second sensor different from the radar, and includes a radar erroneous detection determination unit that determines whether the radar-detected object information obtained from the first detection device is suspected of being an erroneous detection and sets a radar erroneous detection suspicion flag to true for radar-detected object information that is judged to be suspected of being an erroneous detection, and a fusion unit that outputs fusion object information including a fusion object position, which is position information of the fusion object, based on the second-sensor-detected object information and the radar-detected object information obtained from the radar erroneous detection determination unit. The fusion processing unit includes a radar processing unit and a radar false detection deletion unit that deletes fusion object information that is determined to be suspected of being a false detection, and when it is determined that the radar-detected object and the second sensor-detected object are a single fusion object, the fusion processing unit combines the respective radar-detected object information and second sensor-detected object information into a single fusion object information and sets the single sensor detection flag of the combined fusion object information to false, and when it is determined that the fusion object was detected only by radar, it treats the radar-detected object information of the radar-detected object as fusion object information and sets the single sensor detection flag of the fusion object information to true, and the radar false detection deletion unit deletes fusion object information for which the radar false detection suspicion flag is true and the single sensor detection flag is true, thereby suppressing the output of incorrect fusion objects due to false detection due to radar information. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a block diagram showing a configuration of an object detection device in accordance with Embodiment 1. FIG. [Figure 2]FIG. 3 is a diagram showing a radar detection range and a camera detection range in the first embodiment. [Figure 3] 5A to 5C are diagrams illustrating examples of erroneous detection that occur in the first detection device according to the first embodiment. [Figure 4] 10A and 10B are diagrams illustrating another example of erroneous detection that occurs in the first detection device according to the first embodiment. [Figure 5] 4 is a flowchart illustrating the operation of the object detection device according to the first embodiment. [Figure 6] 5 is a flowchart showing details of a radar erroneous detection determination process in a radar erroneous detection determination unit according to the first embodiment. [Figure 7] FIG. 10 is a block diagram showing the configuration of an object detection device according to a second embodiment. [Figure 8] 10 is a flowchart illustrating the operation of the object detection device according to the second embodiment. [Figure 9] 10 is a flowchart showing details of a fusion erroneous detection determination process in a fusion erroneous detection determination unit according to the second embodiment. [Figure 10] FIG. 2 is a schematic diagram illustrating an example of a hardware configuration of an object detection device according to an embodiment. [Figure 11] FIG. 10 is a schematic diagram illustrating another example of the hardware configuration of the object detection device according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an object detection device, an object detection method, and an object detection program according to embodiments of the present invention will be described in detail with reference to the drawings. Note that the same reference numerals in each drawing indicate the same or corresponding parts.

[0012] Embodiment 1 FIG. 1 is a block diagram showing the configuration of an object detection device 100 according to a first embodiment. The object detection device 100 detects an object based on at least information from a first detection device 1 and information from a second detection device 2. FIG. 1 illustrates the object detection device 100 detecting an object based on information from the first detection device 1, information from the second detection device 2, and information from a vehicle information sensor 3. The first detection device 1 is equipped with a radar 11, detects an object based on the output of the radar 11, and outputs radar-detected object information including a radar-detected position, which is position information of the radar-detected object, for the radar-detected object that is an object detected based on the output of the radar 11. The second detection device 2 is equipped with a second sensor different from the radar. The second sensor equipped in the second detection device 2 may be any sensor different from the radar, such as a LiDAR (Light Detection and Ranging) sensor, an ultrasonic sensor (sonar sensor), or an infrared camera. In the following description, the second detection device 2 is described as being equipped with a camera 12. The second detection device 2 includes a camera 12 serving as a second sensor, detects an object based on the output of the camera 12 serving as the second sensor, and outputs second-sensor-detected object information including a second-sensor-detected position, which is position information of the second-sensor-detected object, for the second-sensor-detected object, which is an object detected based on the output of the camera 12 serving as the second sensor. Note that the first detection device 1 and the radar 11 may be the same, and for example, the radar 11 may detect an object and output radar-detected object information including a radar-detected position, which is position information of the radar-detected object. Furthermore, the second detection device 2 and the second sensor may be the same, and for example, the camera 12 serving as the second sensor may detect an object and output second-sensor-detected object information including a second-sensor-detected position, which is position information of the second-sensor-detected object.

[0013] 2 is a diagram showing the detection range of radar 11 and the detection range of camera 12 in embodiment 1. Object detection device 100 is mounted on, for example, vehicle 101, and radar 11 and camera 12 are installed in front of vehicle 101. In FIG. 2, radar detection range 102, which is the detection range of radar 11, is the range indicated by diagonal lines slanting downward to the right, and camera detection range 103, which is the detection range of camera 12, is the range indicated by diagonal lines slanting upward to the right. Radar detection range 102 and camera detection range 103 partially overlap, and the overlapping portion of radar detection range 102 and camera detection range 103 is fusion range 104; in FIG. 2, the overlapping portion of the diagonal lines slanting upward to the right is fusion range 104.

[0014] The object detection device 100 detects an object ahead of the vehicle 101 as a fusion object using output information from the first detection device 1 and output information from the second detection device 2, and outputs the detection result to a subsequent processing block (a control system or an alarm system), thereby realizing autonomous driving or driving assistance. The object detection device 100 may be implemented in a form integrated with other components inside the vehicle 101 or in a form that cannot be separated from them, or may be implemented in a form that can be removed or separated from them. The vehicle information sensor 3 outputs vehicle information such as the traveling speed, yaw rate, azimuth angle, or traveling position of the vehicle 101, and the object detection device 100 may detect an object ahead of the vehicle 101 as a fusion object using the output information from the first detection device 1, the output information from the second detection device 2, and the output information from the vehicle information sensor 3.

[0015] The vehicle 101 is, for example, an automobile, but is not limited to this and may be a four-wheeled vehicle such as a truck or a golf cart, or may be a moving body such as a two-wheeled vehicle, a PMV (Personal Mobility Vehicle), or an AMR (Autonomous Mobile Robot).

[0016] The radar 11 is a sensor that measures the distance and angle to an object by emitting radio waves, for example, in the millimeter wave band, and receiving the radio waves reflected by the object to be detected. The first detection device 1 equipped with the radar 11 detects an object based on the output of the radar 11, and for the detected object, the radar-detected object, calculates radar-detected position information, which is position information of the detected object, based on distance and angle measurement information, and outputs radar-detected object information including the radar-detected position information. The first detection device 1 may also output auxiliary information about the radar-detected object, such as speed, acceleration, reflection intensity (or information such as RCS (Radar Cross-Section) correlated with reflection intensity), size (width, length, etc.), type, and reliability, as radar-detected object information. The radar 11 can detect objects at a distance of 100 m or more, but due to its low lateral resolution, at close distances of, for example, 30 m or less, it becomes difficult to separate nearby objects due to signal processing within the radar ECU (Electronic Control Unit). As a result, a plurality of different objects (or noises) may be clustered into one and output from the first detection device 1, which may result in erroneous detection.

[0017] FIG. 3 is a diagram showing an example of erroneous detection occurring in the first detection device 1 of the first embodiment. Due to the characteristics of the radar 11 described above, noise 105, which is a behavior of something jumping out from the side toward the center of the vehicle while the vehicle is moving, may appear in the output of the first detection device 1, particularly at short distances. FIG. 4 is a diagram showing another example of erroneous detection occurring in the first detection device 1 of the first embodiment. Due to the characteristics of the radar 11 described above, noise 106, which is large in size, may appear in the output of the first detection device 1 while the vehicle is moving, particularly at short distances. Noises such as noise 105 and noise 106 cannot be completely removed by processing in the radar ECU, and it is difficult to reliably distinguish them from actual pedestrians crossing the road, pedestrians jumping out, or actual vehicles moving close by.

[0018] The second detection device 2 detects objects based on the output of the camera 12, and detects objects captured within the viewing angle of the camera 12 by learning images in advance. Detection targets include, for example, people, automobiles, bicycles, motorcycles, etc. The second detection device 2 obtains second-sensor-detected position information, which is position information of the detected object, based on information acquired from the camera 12, and outputs second-sensor-detected object information including the second-sensor-detected position. The camera 12 may be a monocular camera that determines the horizontal position and a rough vertical position, with the main purpose of identifying the type of object, or a compound-eye camera (stereo camera) that determines the vertical position more accurately. The camera 12 may be either a monocular camera or a compound-eye camera.

[0019] The radar erroneous detection determination unit 4 of the object detection device 100 receives radar-detected object information from the first detection device 1 for the number of detected objects, determines whether each piece of radar-detected object information is suspected of being a false detection, sets the radar erroneous detection suspicion flag for radar-detected object information determined to be suspected of being a false detection to true, and sets the radar erroneous detection suspicion flag for radar-detected object information not determined to be suspected of being a false detection to false, and outputs the radar-detected object information. The radar erroneous detection determination unit 4 may determine whether each piece of radar-detected object information is suspected of being a false detection, and add radar erroneous detection information indicating that the radar-detected object information determined to be suspected of being a false detection to the radar-detected object information determined to be suspected of being a false detection, and output the radar-detected object information. Note that the radar erroneous detection determination unit 4 may determine whether each piece of radar-detected object information is suspected of being a false detection based on the radar-detected object information output by the first detection device 1 for the number of detected objects and the vehicle information output by the vehicle information sensor 3. The method for determining false detection by the radar erroneous detection determination unit 4 will be described in detail later.

[0020] The radar filter processing unit 5 performs radar filtering on the radar-detected object information acquired from the radar erroneous detection determination unit 4 and outputs the result after removing unnecessary radar-detected object information. For example, if the range of the positions of objects that will ultimately be controlled or notified is predetermined as a detection target range, the radar filtering process in the radar filter processing unit 5 may remove radar-detected object information whose radar detection position is outside the detection target range. For example, if the radar-detected object information includes information on the detected object velocity, which is the speed of the radar-detected object, the radar filtering process in the radar filter processing unit 5 may remove radar-detected object information whose detected object velocity exceeds a predetermined normal speed range. For example, if the radar-detected object information includes information on the detected object acceleration, which is the acceleration of the radar-detected object, the radar filtering process in the radar filter processing unit 5 may remove radar-detected object information whose detected object acceleration exceeds a predetermined normal acceleration range. Here, the normal speed range and normal acceleration range may be determined based on invalid or abnormal values ​​specified in the specifications of the radar 11, or may be determined depending on the final application method for control or notification. More specifically, the normal speed range and normal acceleration range may delete information about high-speed moving objects that cannot occur in the driving area, even if they are normal values ​​according to the specifications of the output of the radar 11. Furthermore, the radar filtering process in the radar filter processing unit 5 may delete radar-detected object information whose detected object reflection intensity is smaller than a predetermined reflection intensity threshold, for example, when the radar-detected object information includes information about the detected object reflection intensity, which is the reflection intensity of the radar-detected object. The radar filtering process in the radar filter processing unit 5 may be any of the above methods or a combination of multiple above methods. If the radar filtering process in the radar filter processing unit 5 is unnecessary, the radar filtering process in the radar filter processing unit 5 may be skipped, and the fusion processing unit 7 may acquire the radar-detected object information from the radar erroneous detection determination unit 4.

[0021] The camera filter processing unit 6 performs camera filtering on the second-sensor-detected object information acquired from the second detection device 2, and outputs the processed second-sensor-detected object information to the fusion processing unit 7. Similar to the radar filtering process in the radar filtering processing unit 5, the camera filtering process in the camera filtering processing unit 6 may delete second-sensor-detected object information whose second-sensor detection position is outside the detection target range, for example, when the range of positions of objects that will ultimately be subject to control or notification is predetermined as the detection target range. When the camera filtering process in the camera filtering processing unit 6 is unnecessary, the camera filtering process in the camera filtering processing unit 6 may be skipped, and the fusion processing unit 7 may acquire the second-sensor-detected object information from the second detection device 2.

[0022] The fusion processing unit 7 performs sensor fusion processing (hereinafter also referred to simply as "fusion processing") based on the radar-detected object information acquired from the radar filter processing unit 5 and the second-sensor-detected object information acquired from the camera filter processing unit 6, and outputs fusion object information including fusion object position, which is position information of the fusion object, for the fusion object obtained as a result of the fusion processing. If the radar filter processing by the radar filter processing unit 5 and the camera filter processing by the camera filter processing unit 6 are skipped, the fusion processing unit 7 performs fusion processing based on the radar-detected object information acquired from the radar erroneous detection determination unit 4 and the second-sensor-detected object information acquired from the second detection device 2, and outputs fusion object information including fusion object position, which is position information of the fusion object, for the fusion object obtained as a result of the fusion processing.

[0023] The fusion process first compares each piece of radar-detected object information with each piece of second-sensor-detected object information and determines whether the radar-detected object and the second-sensor-detected object are a single fusion object based on the radar detection position and second-sensor-detected position information. For example, the fusion process determines that radar-detected object information and second-sensor-detected object information whose radar detection position and second-sensor-detected position are close to each other are information about the same object, and determines that the radar-detected object indicated by the radar-detected object information and the second-sensor-detected object indicated by the second-sensor-detected object information are a single fusion object. Specifically, the fusion process determines that the radar-detected object and the second-sensor-detected object are a single fusion object when the second-sensor-detected position of the second-sensor-detected object is within a predetermined same-object determination range that includes the radar detection position of the radar-detected object. The fusion process then combines the radar-detected object information and the second-sensor-detected object information into a single fusion object information and sets the single-sensor detection flag of the combined fusion object information to false to indicate that the combined fusion object was detected by both the radar 11 and the second sensor. One method of indicating that the combined fusion object was detected by both the radar 11 and the second sensor may be, for example, to add first detection information indicating that the combined fusion object was detected by both the radar 11 and the second sensor to the fusion object information. When radar-detected object information and second-sensor-detected object information are combined into one piece of fusion object information, information contained only in either the radar-detected object information or the second-sensor-detected object information is carried over to the fusion object information as is. For example, since information on the radar false detection suspicion flag is information contained only in the radar-detected object information, the information on the radar false detection suspicion flag in the radar-detected object information is made information on the radar false detection suspicion flag in the fusion object information.For example, the fusion object information is determined by selecting the information with the higher reliability from the information included in both the radar-detected object information and the second-sensor-detected object information. For example, the information included in the radar-detected object information is selected for vertical position information, and the information included in the second-sensor-detected object information is selected for horizontal position information. Alternatively, rather than selecting one of the pieces of information included in the radar-detected object information and the second-sensor-detected object information, the fusion object information may be determined by weighting the pieces of information included in the radar-detected object information and the second-sensor-detected object information at a predetermined ratio, taking into account the data output cycle of the radar 11 and the camera 12 or missing detections, etc.

[0024] Next, in the fusion processing, when a second sensor-detected object is not detected inside the same object determination range including the radar detection position of the radar-detected object, the fusion processing determines that the fusion object is detected only by the radar 11, sets the radar-detected object information of the radar-detected object as fusion object information, and sets a single sensor detection flag in the fusion object information to true to indicate that the fusion object is detected only by the radar 11. A method of indicating that the combined fusion object is detected only by the radar 11 may, for example, be to add second detection information indicating that the fusion object is detected only by the radar 11 to the fusion object information. If the radar-detected object information does not include information on the type of radar-detected object but the second sensor-detected object information includes information on the type of second sensor-detected object, whether the combined fusion object is detected by both the radar 11 and the second sensor or only by the radar 11 may be indicated by including information on the type of second sensor-detected object in the fusion object information when the combined fusion object is detected by both the radar 11 and the second sensor.

[0025] The determination of whether the second sensor-detected position of the second sensor-detected object is within the same object determination range including the radar detection position of the radar-detected object may be made, for example, by determining that the second sensor-detected position is within the same object determination range when the distance from the radar detection position of the radar-detected object to the second sensor-detected position of the second sensor-detected object is smaller than a predetermined first distance threshold. Specifically, when the radar detection position is expressed as (x1, y1) from x1, the horizontal position, and y1, the second sensor-detected position is expressed as (x2, y2) from x2, the horizontal position, and y2, the vertical position, and the first distance threshold is th1, the second sensor-detected position may be determined to be within the same object determination range when the following formula (1) is satisfied:

[0026]

number

[0027] Alternatively, the determination of whether the second sensor detected position of the second sensor-detected object is inside the same object determination range including the radar detected position of the radar-detected object may be made, for example, by determining that the second sensor detected position is inside the same object determination range when the difference between the horizontal position of the radar detected position of the radar-detected object and the horizontal position of the second sensor detected position of the object detected by the second sensor is smaller than a predetermined horizontal threshold, and the difference between the vertical position of the radar detected position of the radar-detected object and the vertical position of the second sensor detected position of the object detected by the second sensor is smaller than a predetermined vertical threshold. Specifically, the radar detected position is represented by (x1, y1), the second sensor detected position is represented by (x2, y2), and the horizontal threshold is th x and the vertical threshold is th y When the following expressions (2) and (3) are satisfied, it may be determined that the second sensor detection position is inside the same object determination range.

[0028]

number

[0029] In the fusion process, the radar detection position is represented by (x1, y1), the vertical velocity of the radar-detected object, i.e., the velocity in the depth direction, is represented by vy1, the second sensor detection position is represented by (x2, y2), the vertical velocity of the second sensor-detected object, i.e., the velocity in the depth direction, is represented by vy2, the first distance threshold is represented by th1, and the vertical velocity threshold is represented by th vy When the above-described formula (1) is satisfied and the following formula (4) is also satisfied, it may be determined that the radar-detected object and the second-sensor-detected object are one fusion object. In this case, the fusion process may use the radar-detected position of the radar-detected object as a reference and, when a second-sensor-detected object that satisfies formulas (1) and (4) is not detected, determine that the fusion object is detected only by the radar 11.

[0030]

number

[0031] Alternatively, the fusion process may be performed by expressing the radar detection position as (x1, y1), the vertical velocity of the radar-detected object, i.e., the velocity in the depth direction, as vy1, expressing the second sensor detection position as (x2, y2), expressing the vertical velocity of the second sensor-detected object, i.e., the velocity in the depth direction, as vy2, and setting the horizontal threshold as th. x and the vertical threshold is th y and the vertical velocity threshold is th vy When the above-described formulas (2) and (3) are satisfied and the above-described formula (4) is satisfied, the radar-detected object and the second-sensor-detected object may be determined to be one fusion object. In this case, the fusion process may use the radar-detected position of the radar-detected object as a reference and, when a second-sensor-detected object that satisfies formulas (2), (3), and (4) is not detected, determine that the fusion object is detected only by the radar 11.

[0032] Alternatively, the fusion process may be performed by expressing the radar detected position as (x1, y1), expressing the speed of the radar detected object as (vx1, vy1) from vx1, which is the horizontal speed, and vy1, which is the vertical speed, i.e., the speed in the depth direction, and expressing the second sensor detected position as (x2, y2), expressing the speed of the second sensor detected object as (vx2, vy2) from vx2, which is the horizontal speed, and vy2, which is the vertical speed, i.e., the speed in the depth direction, and expressing the first distance threshold as th1, and expressing the speed threshold as th2. v When the above-described formula (1) is satisfied and the following formula (5) is also satisfied, it may be determined that the radar-detected object and the second-sensor-detected object are one fusion object. In this case, the fusion process may use the radar-detected position of the radar-detected object as a reference and, when a second-sensor-detected object that satisfies formulas (1) and (5) is not detected, determine that the fusion object is detected only by the radar 11.

[0033]

number

[0034] Alternatively, the fusion process may be performed by expressing the radar detected position as (x1, y1), the speed of the radar detected object as (vx1, vy1), the second sensor detected position as (x2, y2), the speed of the second sensor detected object as (vx2, vy2), and the lateral threshold as th x and the vertical threshold is th y The third lateral velocity threshold is th vx and the vertical velocity threshold is th vy When the above-described formulas (2), (3), and (4) are satisfied and the following formula (6) is also satisfied, it may be determined that the radar-detected object and the second-sensor-detected object are one fusion object. In this case, the fusion process may use the radar-detected position of the radar-detected object as a reference and determine that the fusion object is detected only by the radar 11 when a second-sensor-detected object that satisfies formulas (2), (3), (4), and (6) is not detected.

[0035]

number

[0036] Alternatively, the fusion process may determine whether the radar-detected object and the second sensor-detected object are a single fusion object by combining the equations (1) to (6).

[0037] The fusion processing unit 7 may perform fusion processing based on the radar-detected object information acquired from the radar filter processing unit 5, the second-sensor-detected object information acquired from the camera filter processing unit 6, and the vehicle information acquired from the vehicle information sensor 3, and output fusion object information including fusion object positions, which are position information of the fusion objects, for the fusion objects obtained as a result of the fusion processing. For example, the fusion processing unit 7 may calculate the ground speed of each radar-detected object and each second-sensor-detected object based on the radar-detected object information, the second-sensor-detected object information, and the vehicle information, and determine whether each radar-detected object and each second-sensor-detected object are stationary or moving. Furthermore, even if the second-sensor-detected position of the second-sensor-detected object is within a predetermined same-object determination range that includes the radar-detected position of the radar-detected object, if one of the radar-detected object and the second-sensor-detected object is stationary and the other is moving, it may be determined that the radar-detected object and the second-sensor-detected object are not a single fusion object. Alternatively, even if the second sensor detection position of the second sensor-detected object is within a predetermined same object determination range that includes the radar detection position of the radar-detected object, if one of the radar-detected object and the second sensor-detected object is a stationary object and the other is a moving object, the threshold value or the like may be changed so that it is less likely that the radar-detected object and the second sensor-detected object will be determined to be a single fusion object.

[0038] When only objects detected by both radar 11 and the second sensor are considered to be fusion objects, objects detected by radar 11 but not by the second sensor, or objects that are not the detection targets of second detection device 2 when the second sensor is camera 12 (objects of a type not previously learned) cannot be detected and output as fusion objects. By considering objects detected only by radar 11 as fusion objects, object detection device 100 according to embodiment 1 can prevent information about objects detected only by radar 11 from being erroneously deleted.

[0039] Depending on the shape, size, and material of the detected object, a single object may be detected as multiple objects by the first detection device 1 or the second detection device 2. Therefore, the fusion processing in the fusion processing unit 7 may perform an integration process to integrate multiple fusion objects whose fusion object positions included in the fusion object information are closer than a predetermined third distance threshold. For example, when the fusion object positions included in the fusion object information are closer than a predetermined third distance threshold, the fusion object information is integrated. When integrating multiple pieces of fusion object information, highly reliable information may be selected from the fusion object information to be used as the newly integrated fusion object information. Alternatively, the information included in each piece of fusion object information may be weighted at a predetermined rate to be used as the newly integrated fusion object information. When integrating multiple pieces of fusion object information, the size of the object including all of the fusion object positions included in each piece of fusion object information may be used as the size information of the newly integrated fusion object information. When the values ​​of all of the radar misdetection suspicion flags of multiple fusion object information to be combined into one are true, the value of the radar misdetection suspicion flag of the newly combined fusion object information may be set to true; when the value of at least one of the radar misdetection suspicion flags of the multiple fusion object information to be combined into one is true, the value of the radar misdetection suspicion flag of the newly combined fusion object information may be set to true; when the values ​​of the radar misdetection suspicion flags of more than half of the multiple fusion object information to be combined into one are true, the value of the radar misdetection suspicion flag of the newly combined fusion object information may be set to true.Furthermore, if the values ​​of all the single sensor detection flags of the multiple fusion object information to be combined are true, the value of the single sensor detection flag of the newly combined fusion object information may be set to true. If even one of the single sensor detection flags of the multiple fusion object information to be combined is false, the value of the single sensor detection flag of the newly combined fusion object information may be set to false. Furthermore, in either case, the obtained fusion object information may be subjected to various post-processing processes, such as tracking or smoothing, to correct the results and improve accuracy or precision. Furthermore, the movement of the vehicle 101 may be predicted based on the vehicle information acquired from the vehicle information sensor 3, and tracking processing may be performed based on the predicted results. For example, because the vehicle 101 moves during times such as sensor processing delays or millimeter wave propagation delays, the values ​​output by the sensors, such as the radar 11 or camera 12, may differ from the actual current positional relationship between the vehicle 101 and the fusion object. Therefore, tracking processing may be performed using values ​​corrected by predicting the error between the sensor output value and the actual current positional relationship between the vehicle 101 and the fusion object.

[0040] For each fusion object obtained by the fusion process, corresponding radar-detected object information is always present. Therefore, the value of the radar false detection suspicion flag in the fusion object information is set to the value of the radar false detection suspicion flag in the corresponding radar-detected object information. In other words, the value of the radar false detection suspicion flag in each radar-detected object information is inherited by the radar false detection suspicion flag in the corresponding fusion object information. Furthermore, if the radar-detected object information includes radar false detection information, the radar false detection information is added to the corresponding fusion object information. Note that if the output cycle of the radar-detected object information from the first detection device 1 differs from the execution cycle of the fusion process by the fusion processing unit 7, there is a possibility that corresponding radar-detected object information does not exist in a certain execution cycle. In this case, the radar false detection information in the fusion object information of the same fusion object in the previous cycle is inherited. Therefore, the value of the corresponding radar false detection suspicion flag is always inherited to all fusion object information.

[0041] The radar false detection removal unit 8 acquires fusion object information from the fusion processing unit 7, removes fusion object information that is determined to be suspected of being a false detection, and outputs the remaining fusion object information that has not been removed to the object selection unit 9. The method by which the radar false detection removal unit 8 determines whether to remove fusion object information will be described in detail later.

[0042] The object selection unit 9 performs object selection processing on the fusion object information obtained from the radar false detection deletion unit 8, and outputs the selection result as the final object detection result. The object selection processing can take, for example, a method of deleting fusion object information including parameters outside the range of normal values. That is, the same processing as that of the radar filter processing unit 5 or the camera filter processing unit 6 can be performed. Also, for the fusion object information, a priority can be calculated based on a predetermined criterion, and a predetermined number of fusion object information can be selected and output in descending order of priority. More specifically, when, due to restrictions such as processing load, there are actually N fusion objects but it is necessary to reduce them to M (<N) objects, for example, if the subsequent control processing block has a requirement to preferentially output nearby objects, the N fusion object information is prioritized in ascending order of distance, and M fusion object information with high priority is selected and output. Also, for example, if there is a requirement to prioritize objects existing on the same lane as the vehicle 101, prioritization can be performed by referring to parameters such as lateral position or lane information in addition to the distance. The object selection processing may be any of the above methods, or a combination of the above methods. Also, if not necessary, the processing in the object selection unit 9 can be skipped, and the output of the radar false detection deletion unit 8 can be output as the final object detection result.

[0043] FIG. 5 is a flowchart for explaining the operation of the object detection device 100 according to the first embodiment. Step S01 is a radar false detection determination step for performing radar false detection determination processing, step S02 is a radar filter processing step for performing radar filter processing, step S03 is a camera filter processing step for performing camera filter processing, step S04 is a fusion processing step for performing fusion processing, steps S05, S06, and S07 are radar false detection deletion steps for performing radar false detection deletion, and step S08 is an object selection step for performing object selection.

[0044] In step S01, the radar erroneous detection determination unit 4 determines whether each piece of radar-detected object information acquired from the first detection device 1 is suspected of being a false detection, sets the radar erroneous detection suspicion flag of radar-detected object information determined to be suspected of being a false detection to true, sets the radar erroneous detection suspicion flag of radar-detected object information determined not to be suspected of being a false detection to false, outputs the radar-detected object information to the radar filter processing unit 5, and proceeds to step S02. The method of determining false detection will be explained in detail later.

[0045] In step S02, the radar filter processing unit 5 performs radar filter processing on each piece of radar-detected object information acquired from the radar erroneous detection determination unit 4, outputs the processed radar-detected object information to the fusion processing unit 7, and proceeds to step S03. In step S03, the camera filter processing unit 6 performs camera filter processing on the second-sensor-detected object information acquired from the second detection device 2, outputs the processed second-sensor-detected object information to the fusion processing unit 7, and proceeds to step S04.

[0046] In step S04, the fusion processing unit 7 performs fusion processing based on the radar-detected object information acquired from the radar filter processing unit 5 and the second-sensor-detected object information acquired from the camera filter processing unit 6, and outputs fusion object information including fusion object positions, which are position information of the fusion objects, for each fusion object obtained as a result of the fusion processing to the radar erroneous detection removal unit 8, and then proceeds to step S05. Note that steps S02 and S03 may be skipped, and if steps S02 and S03 are skipped, the fusion processing unit 7 performs fusion processing based on the radar-detected object information acquired from the radar erroneous detection determination unit 4 and the second-sensor-detected object information acquired from the second detection device 2. In the fusion processing, for example, the value of the radar erroneous detection suspicion flag is transferred from the radar-detected object information to the fusion object information.

[0047] In step S05, the radar false detection elimination unit 8 checks the value of the radar false detection suspicion flag of each piece of fusion object information acquired from the fusion processing unit 7, and proceeds to step S06 if the radar false detection suspicion flag is true, and proceeds to step S08 if the radar false detection suspicion flag is false. In step S06, the radar false detection elimination unit 8 checks the value of the single sensor detection flag of each fusion object information, and proceeds to step S07 if the value of the single sensor detection flag is true, and proceeds to step S08 if the value of the single sensor detection flag is false. In step S07, the radar false detection elimination unit 8 deletes the fusion object information and proceeds to step S08. By the processes from step S05 to step S07, the radar false detection elimination unit 8 deletes fusion object information for which the radar false detection suspicion flag is true and the single sensor detection flag is true, and outputs fusion object information for which only one of the radar false detection suspicion flag and the single sensor detection flag is true or for which both the radar false detection suspicion flag and the single sensor detection flag are false to the object selection unit 9. That is, the radar erroneous detection deletion unit 8 deletes the fusion object information when the fusion object is detected only by the radar 11 and the detection by the radar 11 is suspected to be an erroneous detection.

[0048] In step S08, the object selection unit 9 performs object selection processing on each piece of fusion object information acquired from the radar false detection elimination unit 8, and outputs the selection result as the final object detection result. Note that step S08 may be skipped, and in that case, the output of the radar false detection elimination unit 8 is output as the final object detection result.

[0049] Fig. 6 is a flowchart for explaining the content of the radar erroneous detection determination process in the radar erroneous detection determination unit 4 of the first embodiment, and shows details of the radar erroneous detection determination process shown in step S01 of Fig. 5. Step S11 is a first erroneous detection determination step, steps S12 and S13 are first erroneous detection flag setting steps, step S14 is a second erroneous detection determination step, steps S15 and S16 are second erroneous detection flag setting steps, step S17 is a first erroneous detection flag checking step, step S18 is a second erroneous detection flag checking step, and steps S19 and S20 are radar erroneous detection suspicion flag setting steps.

[0050] In step S11, the radar erroneous detection determination unit 4 performs a first erroneous detection determination to determine whether the content of each piece of radar-detected object information acquired from the first detection device 1 satisfies the first erroneous detection condition. The first erroneous detection determination is performed by checking whether the content of the radar-detected object information satisfies the first erroneous detection condition, and if the content of the radar-detected object information satisfies the first erroneous detection condition, the process proceeds to step S12, and if the content of the radar-detected object information does not satisfy the first erroneous detection condition, the process proceeds to step S13.

[0051] As described above, noise 105, which appears as if it is projecting from the side toward the center of the vehicle, may appear in the output of the first detection device 1, as shown in FIG. 3. Therefore, radar-detected object information exhibiting behavior such as that shown by noise 105 in FIG. 3 is determined to be due to a first erroneous detection. To determine whether a first erroneous detection has occurred, information indicating the behavior of the radar-detected object contained in the radar-detected object information is used. Here, the information indicating the behavior of the radar-detected object is expressed as time-series information for at least one of the radar-detected position (ordinate and abscissa), which is position information of the radar-detected object, the moving speed (longitudinal speed and lateral speed) of the radar-detected object, and the moving acceleration (longitudinal acceleration and lateral acceleration) of the radar-detected object. Alternatively, information equivalent to these may be calculated and used instead. For example, the result of a differential calculation (difference calculation) of the radar-detected position may be used instead of the moving speed of the radar-detected object. More specifically, the information indicating the behavior of the radar-detected object may be, for example, "changes in the values ​​of the current radar detection position, the radar detection position one processing cycle ago, and the radar detection position two processing cycles ago" or "changes in the values ​​of the current lateral velocity, the lateral velocity one processing cycle ago, and the lateral velocity two processing cycles ago." The first erroneous detection determination is performed by determining whether the information indicating the behavior of the radar-detected object is a behavior of the radar-detected object jumping out from the side toward the vehicle center axis (lateral position = 0). Specifically, for example, the radar erroneous detection determination unit 4 determines that the first erroneous detection has occurred when the amount of change in the radar detection position of the radar-detected object moving toward the center in the lateral direction in the radar detection range 102, which is the detection range of the radar 11, is greater than a predetermined first change amount threshold continuously over a predetermined first observation time (for example, the time of three processing cycles). That is, the radar erroneous detection determination unit 4 determines that a first erroneous detection has occurred when the value of the speed at which the radar-detected object moves toward the center in the horizontal direction of the radar detection range 102 is greater than a predetermined first change amount threshold continuously over a predetermined first observation time. Furthermore, when the radar-detected object information includes information on a first lateral velocity that is the lateral movement speed of the radar-detected object, the radar erroneous detection determination unit 4 may determine that a first erroneous detection has occurred when the absolute value of the first lateral velocity is greater than a predetermined first lateral velocity threshold continuously over the first observation time.

[0052] Furthermore, the determination of the first erroneous detection may be limited to a certain range, such as by determining only radar-detected objects that are located in close proximity, rather than determining the entire fusion range 104. For example, the threshold value may be changed according to the position, speed, or acceleration of the radar-detected object. For example, the position, speed, or acceleration of the radar-detected object may be determined by using radar-detected object information acquired from the first detection device 1 and vehicle information acquired from the vehicle information sensor 3 to perform a path coordinate transformation according to the inclination of the vehicle 101. Similarly, for the speed of the radar-detected object, instead of the moving speed of the radar-detected object in the radar-detected object information (relative speed with respect to the vehicle 101), a value obtained by converting the moving speed of the radar-detected object into a ground speed using the radar-detected object information acquired from the first detection device 1 and the vehicle information acquired from the vehicle information sensor 3 may be used. For example, once a radar-detected object has been determined to be a first erroneous detection, the determination result of the first erroneous detection may be maintained for the same object for a certain period, or until the radar-detected object determined to be a first erroneous detection is no longer detected.

[0053] In step S12, the radar erroneous detection determination unit 4 sets the value of the first erroneous detection flag of the radar-detected object information to true, and the process proceeds to step S14. In step S13, the radar erroneous detection determination unit 4 sets the value of the first erroneous detection flag of the radar-detected object information to false, and the process proceeds to step S14.

[0054] In step S14, a second erroneous detection determination is performed to determine whether the content of each piece of radar-detected object information acquired from the first detection device 1 satisfies the second erroneous detection condition. The second erroneous detection determination is performed by checking whether the content of the radar-detected object information satisfies the second erroneous detection condition. If the content of the radar-detected object information satisfies the second erroneous detection condition, the process proceeds to step S15. If the content of the radar-detected object information does not satisfy the second erroneous detection condition, the process proceeds to step S16. As described above, large noise 106 may appear in the output of the first detection device 1 at close ranges, as shown in FIG. 4. Therefore, radar-detected object information exhibiting a state such as the noise 106 in FIG. 4 is determined to be due to a second erroneous detection. The second erroneous detection determination is performed using information indicating the characteristics of the radar-detected object contained in the radar-detected object information. Here, the information indicating the characteristics of the radar-detected object is expressed as current information or time-series information regarding at least one of the radar-detected object's reflection intensity (or RCS), size (width, length, and height), type, radar detection position (ordinate and abscissa), and reliability. More specifically, for example, the radar erroneous detection determination unit 4 determines that the second erroneous detection has occurred when the radar-detected object information includes information on the size of the radar-detected object and information on the reflection intensity of the radar-detected object, and the size of the radar-detected object is greater than a predetermined size threshold and the reflection intensity of the radar-detected object is less than a predetermined reflection intensity threshold.Further, for example, the radar erroneous detection determination unit 4 determines that the second erroneous detection has occurred when the radar-detected object information includes information on the reflection intensity of the radar-detected object, the distance from the radar 11 to the radar detection position of the radar-detected object is less than a predetermined second distance threshold (for example, when the radar detection position is within a predefined target range in which the vehicle 101 executes control), and the reflection intensity of the radar-detected object is less than a predetermined reflection intensity threshold. Furthermore, for example, the radar erroneous detection determination unit 4 determines that a second erroneous detection has occurred when the radar-detected object information includes information on the reliability of the radar-detected object and information on the reflection intensity of the radar-detected object, and the reliability of the radar-detected object is smaller than a predetermined reliability threshold and the reflection intensity of the radar-detected object is smaller than a predetermined reflection intensity threshold.

[0055] Furthermore, the determination of the second erroneous detection may be limited to a specific range, such as by determining only radar-detected objects present in close proximity, rather than determining the entire fusion range 104. For example, if the radar-detected object information output by the first detection device 1 includes type information, the determination may be performed based on the type information instead of the size information of the radar-detected object. More specifically, instead of confirming that the size of the radar-detected object is larger than a predetermined size threshold, the radar-detected object may be determined to satisfy the size condition if the type of the radar-detected object corresponds to a large object such as a car, truck, or wide object. For example, time-series information indicating the characteristics of the radar-detected object may be used to relax the determination condition for a radar-detected object that suddenly appears, suddenly appears in close proximity, or suddenly appears near the front, making it more likely to be determined to be a second erroneous detection. This is because if there is an actual object, it would not actually appear suddenly, but would generally be first detected at a distance or in a lateral position near the radar viewing angle, and then move closer or in front of the object; therefore, an object that suddenly appears (particularly an object at close range or a large object) is likely to be a second erroneous detection. Furthermore, for example, the position of a radar-detected object may be determined using the result of performing a route coordinate transformation in accordance with the inclination of the vehicle 101, using radar-detected object information acquired from the first detection device 1 and vehicle information acquired from the vehicle information sensor 3. Furthermore, for example, once a radar-detected object has been determined to be a second erroneous detection, the determination result of second erroneous detection may be maintained for the same object for a certain period, or until the radar-detected object determined to be a second erroneous detection is no longer detected.

[0056] In step S15, the radar erroneous detection determination unit 4 sets the value of the second erroneous detection flag of the radar-detected object information to true, and then proceeds to step S17. In step S16, the radar erroneous detection determination unit 4 sets the value of the second erroneous detection flag of the radar-detected object information to false, and then proceeds to step S17.

[0057] In step S17, the radar erroneous detection determination unit 4 checks whether the first erroneous detection flag of each radar-detected object information is true, and if the first erroneous detection flag is true, proceeds to step S19. If the first erroneous detection flag is not true, i.e., if the first erroneous detection flag is false, proceeds to step S18. In step S18, the radar erroneous detection determination unit 4 checks whether the second erroneous detection flag of each radar-detected object information is true, and if the second erroneous detection flag is true, proceeds to step S19. If the second erroneous detection flag is not true, i.e., if the second erroneous detection flag is false, proceeds to step S20. In step S19, the radar erroneous detection determination unit 4 sets the value of the radar erroneous detection suspicion flag of the radar-detected object information to true, and ends the radar erroneous detection determination process. In step S20, the radar erroneous detection determination unit 4 sets the value of the radar erroneous detection suspicion flag of the radar-detected object information to false, and ends the radar erroneous detection determination process.

[0058] The order of the processing related to the first erroneous detection performed in steps S11 to S13 and the processing related to the second erroneous detection performed in steps S14 to S16 may be reversed. Only one of the processing related to the first erroneous detection and the processing related to the second erroneous detection may be performed. When only one of the processing related to the first erroneous detection and the processing related to the second erroneous detection is performed, the value of the flag obtained in the performed processing is used as the value of the radar erroneous detection suspicion flag.

[0059] The radar erroneous detection determination unit 4 may acquire information on the traveling speed, yaw rate, azimuth angle, or traveling position of the vehicle 101 as vehicle information from the vehicle information sensor 3, and may change at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold according to the vehicle information in determining the first erroneous detection and the second erroneous detection. When the vehicle 101 is traveling at high speed, there is a higher possibility that a real object (not noise) will appear at a high relative speed or that a real object will suddenly appear in front of or near the vehicle 101, compared to when the vehicle 101 is traveling at low speed. Furthermore, the faster the traveling speed of the vehicle 101, the higher the risk to the occupants of not performing brake control or notifying for a real object, compared to the risk of erroneous brake control or notifying due to inability to remove noise. Therefore, the radar false detection judgment unit 4 acquires the vehicle speed information of the vehicle 101 from the vehicle information sensor 3, and changes at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold so that the faster the vehicle speed of the vehicle 101, the less likely it is that the radar-detected object information will be judged as a false detection (i.e., so that the fusion object information is not deleted as much as possible), thereby making it possible to make judgments that pose less risk to the occupants.

[0060] Furthermore, the radar erroneous detection determination unit 4 may acquire external environment information including at least one of weather, time, temperature, road surface condition, and the installation status of surrounding structures (such as the installation status of guardrails and the spacing between trees), and may change at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold in accordance with the external environment information when determining the first erroneous detection and the second erroneous detection. For example, at night, the detection performance of the camera 12 significantly deteriorates. Therefore, when the radar erroneous detection suspicion flag is set to true, the probability that the fusion object information will be deleted increases. At night, for the occupants, the risk of not performing braking control or notifying a real object is higher than the risk of erroneous braking control or notifying a real object due to inability to remove noise. Therefore, the radar erroneous detection determination unit 4 may acquire time information and change at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold so that radar-detected object information is less likely to be determined to be an erroneous detection at night (i.e., so that fusion object information is less likely to be deleted). Furthermore, since the detection performance of the camera 12 is reduced during rainy or snowy weather compared to normal conditions, the radar erroneous detection determination unit 4 may acquire weather information and change at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold so that radar-detected object information is less likely to be determined to be an erroneous detection during rainy or snowy weather (i.e., so that fusion object information is less likely to be deleted). Furthermore, when the temperature is low and the road surface is icy, even if the sensor's detection performance is equivalent to normal conditions, the braking distance required for braking or deceleration becomes longer, making it necessary to perform control for obstacles that are further away. Therefore, the radar erroneous detection determination unit 4 may acquire temperature information and change the respective threshold values ​​so that when the temperature is low, it is more likely that a more distant object will be determined to be an erroneous detection.For example, the radar false detection determination unit 4 may acquire temperature information and, when the temperature is higher than a predetermined temperature threshold, perform a first or second false detection determination only for radar-detected objects located within the short distance of the fusion range 104. When the temperature is equal to or lower than the temperature threshold, perform a first or second false detection determination for radar-detected objects located within a wider range. This may make it easier to determine a false detection for objects located further away. For example, if guardrails or trees are installed near the roadway, and the guardrail poles or trees are closely spaced, processing within the radar ECU may cluster multiple objects into one, making it easier for a large object to appear. Generally, guardrails or trees are not located on the roadway and are not objects subject to control or notification. Therefore, these should be eliminated as radar false detections (even though they are real objects). Therefore, when information indicating the presence of a relevant structure is acquired, various thresholds, such as a size threshold, may be changed in that area, for example, so that the clustered structures are more likely to be determined as false detections in the second false detection determination.

[0061] The radar erroneous detection determination unit 4 acquires map information and traveling position information from the vehicle information sensor 3, and may change at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold in accordance with the map information and the traveling position information when determining whether there is a first erroneous detection or a second erroneous detection. The map information is information acquired in advance or recorded, and may be a general map of the area in which the vehicle 101 travels, or high-precision map information including information on the installation status of surrounding structures (such as the installation status of guardrails or the spacing between trees). For example, if it is known through a prior survey, a preliminary drive, or an operation check that radar erroneous detection is likely to occur in a specific area on the map, at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold may be changed when the traveling position is within the specific area so that radar-detected object information is more likely to be determined to be an erroneous detection. Furthermore, for example, when there are many obstacles near the vehicle 101, such as when the vehicle is traveling on a narrow road, it may become difficult to separate nearby objects due to signal processing within the first detection device 1, and noise may be more likely to occur. Therefore, when the vehicle is traveling on a narrow road, at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold may be changed so that radar-detected object information is more likely to be determined as a false detection.

[0062] As described above, object detection device 100 according to embodiment 1 is object detection device 100 that detects objects based on information from first detection device 1 that outputs radar-detected object information including radar-detected position, which is position information of the radar-detected object, for radar-detected objects that are objects detected based on the output of radar 11, and information from second detection device 2 that outputs second-sensor-detected object information including second-sensor-detected position, which is position information of the second-sensor-detected object, for objects detected based on the output of a second sensor different from radar 11. Object detection device 100 includes radar erroneous detection determination unit 4 that determines whether the radar-detected object information acquired from first detection device 1 is suspected of being an erroneous detection, and sets a radar erroneous detection suspicion flag to true for radar-detected object information that is judged to be suspected of being an erroneous detection, and outputs fusion object information including fusion object position, which is position information of the fusion object, based on the second-sensor-detected object information and the radar-detected object information acquired from radar erroneous detection determination unit 4. and a radar erroneous detection deletion unit 8 that deletes fusion object information determined to be suspected of being a false detection. When the fusion processing unit 7 determines that the radar-detected object and the second-sensor-detected object are a single fusion object, it combines the radar-detected object information and the second-sensor-detected object information into a single fusion object information and sets the single-sensor detection flag of the combined fusion object information to false. When it determines that the fusion object was detected only by the radar 11, it treats the radar-detected object information of the radar-detected object as fusion object information and sets the single-sensor detection flag of the fusion object information to true. The radar erroneous detection deletion unit 8 deletes fusion object information for which the radar erroneous detection suspicion flag is true and the single-sensor detection flag is true, thereby suppressing the output of erroneous fusion objects due to false detections caused by information from the radar 11. Furthermore, by targeting the fusion object information of fusion objects detected only by the radar 11 for deletion, it is possible to prevent erroneous deletion of fusion object information.

[0063] Embodiment 2 Fig. 7 is a block diagram showing the configuration of object detection device 100a according to embodiment 2. Comparing object detection device 100a according to embodiment 2 shown in Fig. 7 with object detection device 100 according to embodiment 1 shown in Fig. 1, object detection device 100a according to embodiment 2 has been added with fusion false detection determination unit 10, fusion processing unit 7 has become fusion processing unit 7a, and radar false detection elimination unit 8 has become radar false detection elimination unit 8a. Other configurations of object detection device 100a according to embodiment 2 are the same as those of object detection device 100 according to embodiment 1.

[0064] The fusion processing unit 7a performs the same operations as the fusion processing unit 7, and also performs tracking processing or integration processing in which multiple fusion objects whose fusion object positions included in each fusion object information are closer than a predetermined third distance threshold are combined into one and output as new fusion object information.

[0065] The fusion false detection determination unit 10 acquires fusion object information from the fusion processing unit 7a, determines whether each piece of fusion object information is suspected of being a false detection, sets the fusion false detection suspicion flag of fusion object information determined to be suspected of being a false detection to true, and sets the fusion false detection suspicion flag of fusion object information not determined to be suspected of being a false detection to false, and outputs the fusion object information. The fusion false detection determination unit 10 may determine whether each piece of fusion object information is suspected of being a false detection, and add fusion false detection information indicating that the fusion object information determined to be suspected of being a false detection to the fusion object information determined to be suspected of being a false detection, and output the fusion object information. Note that the fusion false detection determination unit 10 may determine whether each piece of fusion object information is suspected of being a false detection based on the fusion object information output from the fusion processing unit 7a and the vehicle information output from the vehicle information sensor 3. The method of determining false detection by the fusion false detection determination unit 10 will be described in detail later.

[0066] The radar erroneous detection deletion unit 8a acquires fusion object information from the fusion erroneous detection determination unit 10, deletes fusion object information determined to be suspected of being erroneously detected, and outputs the remaining fusion object information that has not been deleted to the object selection unit 9. The method by which the radar erroneous detection deletion unit 8a determines whether to delete fusion object information will be described in detail later.

[0067] Figure 8 is a flowchart illustrating the operation of object detection device 100a according to embodiment 2. In Figure 8, the processes of steps S01 to S03, step S05, and steps S06 to S08 are the same as the processes of steps S01 to S03, step S05, and steps S06 to S08, respectively, of object detection device 100 according to embodiment 1 shown in Figure 5. Step S04a is a fusion processing step that performs fusion processing, step S09 is a fusion erroneous detection determination processing step that performs fusion erroneous detection determination processing, and steps S05, S10, S06, and S07 are radar erroneous detection elimination steps that eliminate radar erroneous detections.

[0068] In step S04a, the fusion processing unit 7a performs the same processing as the fusion processing unit 7, and further performs an integration process to combine multiple fusion objects whose fusion object positions included in each fusion object information are closer than a predetermined third distance threshold and output the combined fusion object information as new fusion object information, and for each resulting fusion object, outputs fusion object information including the fusion object position, which is the position information of the fusion object, to the fusion false detection judgment unit 10, and proceeds to step S09.

[0069] In step S09, the fusion false detection determination unit 10 determines whether each piece of fusion object information acquired from the fusion processing unit 7a is suspected of being a false detection, sets the fusion false detection suspicion flag of the fusion object information that is determined to be a suspected false detection to true, sets the fusion false detection suspicion flag of the fusion object information that is not determined to be a suspected false detection to false, outputs the fusion object information to the radar false detection deletion unit 8a, and proceeds to step S05. The method of determining false detection will be explained in detail later.

[0070] In step S05, the radar false detection elimination unit 8a checks the value of the radar false detection suspicion flag of each piece of fusion object information acquired from the fusion processing unit 7a. If the radar false detection suspicion flag is true, the process proceeds to step S06. If the radar false detection suspicion flag is false, the process proceeds to step S10. In step S10, the radar false detection elimination unit 8a checks the value of the fusion false detection suspicion flag of each piece of fusion object information acquired from the fusion processing unit 7a. If the fusion false detection suspicion flag is true, the process proceeds to step S06. If the fusion false detection suspicion flag is false, the process proceeds to step S08. In step S06, the radar false detection elimination unit 8a checks the value of the single sensor detection flag of each piece of fusion object information. If the value of the single sensor detection flag is true, the process proceeds to step S07. If the value of the single sensor detection flag is false, the process proceeds to step S08. In step S07, the radar false detection elimination unit 8a deletes the fusion object information, and then proceeds to step S08. Through the processes of steps S05 to S07 and step S10, the radar erroneous detection deletion unit 8a deletes fusion object information for which the radar erroneous detection suspicion flag is true and the single sensor detection flag is true, and fusion object information for which the fusion erroneous detection suspicion flag is true and the single sensor detection flag is true, and outputs the remaining fusion object information to the object selection unit 9. That is, the radar erroneous detection deletion unit 8 deletes fusion object information when a fusion object is detected only by the radar 11 and the detection by the radar 11 is suspected to be a erroneous detection, and deletes fusion object information when a fusion object is detected only by the radar 11 and the behavior as a fusion object is the same as that of a fusion object suspected to be a erroneous detection by the radar 11.

[0071] In step S08, the object selector 9 performs object selection processing on each piece of fusion object information acquired from the radar false detection elimination unit 8a, and outputs the selection result as the final object detection result. Note that step S08 may be skipped, and in that case, the output of the radar false detection elimination unit 8a is output as the final object detection result.

[0072] Fig. 9 is a flowchart for explaining the contents of the fusion false detection determination process in the fusion false detection determination unit 10 of the second embodiment, and shows details of the fusion false detection determination process shown in step S09 of Fig. 8. Step S21 is a fusion false detection determination step, and steps S22 and S23 are fusion false detection suspicion flag setting steps.

[0073] In step S21, the fusion false detection determination unit 10 performs a fusion false detection determination to determine whether the content of each fusion object information acquired from the fusion processing unit 7a satisfies the condition for fusion false detection. The fusion false detection determination is performed by checking whether the content of the fusion object information satisfies the condition for fusion false detection. If the content of the fusion object information satisfies the condition for fusion false detection, the process proceeds to step S22, and if the content of the fusion object information does not satisfy the condition for fusion false detection, the process proceeds to step S23.

[0074] The condition for fusion false detection may be the same as the condition for the first false detection in the radar false detection determination unit 4. For example, fusion object information exhibiting behavior such as that shown by noise 105 in FIG. 3 is determined to be due to fusion false detection. Here, the information indicating the behavior of the fusion object is expressed as time-series information for at least one of the fusion object's position information (ordinate and abscissa), the fusion object's movement speed (longitudinal and lateral speed), and the fusion object's movement acceleration (longitudinal and lateral acceleration). Alternatively, information equivalent to these may be calculated and used. For example, the result of differential calculation (difference calculation) of the fusion object's position may be used instead of the fusion object's movement speed. More specifically, the information indicating the behavior of the fusion object may be, for example, "the transitions in the values ​​of the current fusion object position, the fusion object position one processing cycle ago, and the fusion object position two processing cycles ago" or "the transitions in the values ​​of the lateral speed of the current fusion object, the fusion object's lateral speed one processing cycle ago, and the fusion object's lateral speed two processing cycles ago." The determination of fusion false detection is made by determining whether the information indicating the behavior of the fusion object indicates a behavior of jumping out from the side toward the vehicle center axis (lateral position = 0). Specifically, for example, the fusion false detection determination unit 10 determines that the fusion false detection has occurred when the amount of change in the fusion object position as the fusion object moves toward the center in the left-right direction in the radar detection range 102, which is the detection range of the radar 11, is greater than a predetermined second change amount threshold continuously over a predetermined second observation time (for example, the time of three processing cycles). In other words, the fusion false detection determination unit 10 determines that the fusion false detection has occurred when the value of the speed at which the fusion object moves toward the center in the left-right direction in the radar detection range 102 is greater than a predetermined second change amount threshold continuously over the predetermined second observation time.Furthermore, when the fusion object information includes information on a second lateral velocity, which is the lateral movement velocity of the fusion object, the fusion erroneous detection determination unit 10 may determine that a fusion erroneous detection has occurred if the absolute value of the second lateral velocity is continuously greater than a predetermined second lateral velocity threshold over a second observation time. Here, the second observation time may be the same as the first observation time used in the radar erroneous detection determination by the radar erroneous detection determination unit 4, the second change amount threshold may be the same as the first change amount threshold used in the radar erroneous detection determination by the radar erroneous detection determination unit 4, and the second lateral velocity threshold may be the same as the first lateral velocity threshold used in the radar erroneous detection determination by the radar erroneous detection determination unit 4.

[0075] In step S22, the fusion erroneous detection determination unit 10 sets the value of the fusion erroneous detection suspicion flag in the fusion object information to true, and ends the fusion erroneous detection determination process.In step S23, the fusion erroneous detection determination unit 10 sets the value of the fusion erroneous detection suspicion flag in the fusion object information to false, and ends the fusion erroneous detection determination process.

[0076] The fusion erroneous detection determination unit 10 may acquire information on the traveling speed, yaw rate, azimuth angle, or traveling position of the vehicle 101 as vehicle information from the vehicle information sensor 3, and may change at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold in accordance with the vehicle information in determining whether a fusion erroneous detection has occurred. For example, the fusion erroneous detection determination unit 10 may acquire vehicle speed information of the vehicle 101 from the vehicle information sensor 3, and change at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold so that the faster the vehicle speed of the vehicle 101, the less likely it is that the fusion object information will be determined to be an erroneous detection (i.e., so that the fusion object information is not deleted as much as possible), thereby making it possible to make a determination with less risk to the occupants.

[0077] The fusion false detection determination unit 10 may also acquire external environmental information including at least one of weather, time, temperature, road surface condition, and the installation status of surrounding structures (such as the installation status of guardrails and the spacing between trees), and in determining whether a fusion false detection has occurred, change at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold in accordance with the external environmental information. For example, the fusion false detection determination unit 10 may acquire time information and change at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold so that fusion object information is less likely to be determined to be a false detection at night (i.e., so that fusion object information is less likely to be deleted). Furthermore, because the detection performance of the camera 12 is reduced during rainy or snowy weather compared to normal conditions, the fusion false detection determination unit 10 may acquire weather information and change at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold so that fusion object information is less likely to be determined to be a false detection during rainy or snowy weather (i.e., so that fusion object information is less likely to be deleted). The fusion false detection determination unit 10 may also acquire temperature information and change the thresholds so that, at low temperatures, it is more likely to determine that even more distant objects are suspected of being falsely detected. For example, if guardrails or trees are installed near the roadway, and the guardrail poles or trees are installed closely together, multiple objects may be clustered together, making it more likely that a large object will appear. Generally, guardrails or trees are not located on the roadway and are not objects subject to control or notification. Therefore, these should be eliminated as fusion false detections (although they are real objects). Therefore, when information indicating the presence of a relevant structure is acquired, various thresholds, such as a size threshold, may be changed in that area so that the fusion false detection determination unit 10 is more likely to determine that the clustered structures are falsely detected.

[0078] The fusion false detection determination unit 10 may acquire map information and traveling position information from the vehicle information sensor 3, and in determining whether a fusion false detection has occurred, may change at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold according to the map information and the traveling position information. For example, if it has been determined through a prior investigation, a preliminary drive, or an operation check that radar false detection is likely to occur in a specific area on the map, at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold may be changed when the traveling position is within the specific area so that the fusion object information is more likely to be determined to be a false detection. Furthermore, if the traveling position is on a narrow road, at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold may be changed so that the fusion object information is more likely to be determined to be a false detection.

[0079] As described above, the object detection device 100a according to the second embodiment includes a fusion erroneous detection determination unit 10 that determines whether the fusion object information acquired from the fusion processing unit 7a is suspected of being a false detection, and sets the fusion erroneous detection suspicion flag of the fusion object information that is judged to be suspected of being a false detection to true. The fusion processing unit 7a combines multiple pieces of fusion object information whose fusion object positions included in each piece of fusion object information are closer than a predetermined third distance threshold into one piece and outputs the combined piece as new fusion object information. The fusion erroneous detection determination unit 10 determines whether the amount of change in the fusion object position when the fusion object moves toward the center in the left-right direction within the detection range of the radar 11 is predetermined. If the change amount threshold is greater than a predetermined second threshold value over the second observation time, or if the fusion object information includes information on a second lateral velocity, which is the lateral movement velocity of the fusion object, and the absolute value of the second lateral velocity is greater than a predetermined second lateral velocity threshold value over the second observation time, the fusion object information is determined to be suspected of being a false detection, and the radar false detection deletion unit 8a deletes fusion object information for which the radar false detection suspicion flag is true and the single sensor detection flag is true, and deletes fusion object information for which the fusion false detection suspicion flag is true and the single sensor detection flag is true, thereby further suppressing the output of false fusion objects due to false detection due to information from the radar 11.

[0080] As a result of the fusion process, multiple objects are merged, and various correction processes such as tracking are performed, so the values ​​of various parameters, such as position or speed, are generally different between the radar-detected object information and the fusion object information. Therefore, even if the radar-detected object information is not determined to be a suspected erroneous detection, the fusion object may behave similarly to noise. In such cases, the object detection device 100a described in the second embodiment can further suppress the output of erroneous fusion objects due to erroneous detection based on the information from the radar 11.

[0081] In the embodiment, as shown in Fig. 2, a case has been described in which two sensors, a radar 11 and a camera 12 serving as a second sensor, are arranged in front of the vehicle 101, but the arrangement of the sensors is not limited to the front, and they may be arranged at a location other than the front, such as the rear or side. The radar 11 and the second sensor may be installed on different surfaces of the vehicle 101, as long as their respective detection ranges partially overlap, that is, as long as there is a fusion range. The second sensor may also be provided in a roadside unit (infrastructure sensor).

[0082] The object detection device according to the embodiment is characterized in that even if it is determined that information about an object detected by radar 11 is suspected of being a false detection, if the second sensor detects the same object, the object information is not determined to be suspected of being a false detection. Therefore, the second sensor only needs to have object detection characteristics (in other words, a tendency toward false detection) that are different from those of radar 11. Therefore, the second sensor may be a radar that uses a different frequency from radar 11. Furthermore, the second sensor may be the same sensor as radar 11, and the detection signal processing in the second detection device 2 may be different from the detection signal processing in the first detection device 1.

[0083] Furthermore, the second sensor may be a combination of multiple sensors. For example, the radar 11, a camera, and an ultrasonic sensor may be disposed in front of the vehicle 101, and the camera may be used as the second sensor in a fusion range from short distances to long distances of the vehicle 101, and the ultrasonic sensor may be used as the second sensor in a fusion range of a very close distance of the vehicle 101. Alternatively, for example, the radar 11, a camera, and a LiDAR may be disposed in front of the vehicle 101, and the camera and LiDAR may be used as the second sensors, and an object detected by at least one of the camera and the LiDAR may be used as the second sensor-detected object, and the single-sensor detection flag may be set to false when the second sensor-detected position of the second sensor-detected object is within a predetermined same object determination range that includes the radar-detected position of the radar-detected object. Alternatively, for example, the radar 11, camera, and LiDAR may be placed in front of the vehicle 101, the camera and LiDAR may be used as second sensors, an object detected by both the camera and LiDAR may be used as the second sensor-detected object, and the single sensor detection flag may be set to false when the second sensor-detected position of the second sensor-detected object is within a predetermined same object determination range that includes the radar-detected position of the radar-detected object.

[0084] FIG. 10 is a schematic diagram illustrating an example of the hardware configuration of the object detection device 100, 100a according to the embodiment. The radar false detection determination unit 4, radar filter processing unit 5, camera filter processing unit 6, fusion processing units 7, 7a, radar false detection elimination units 8, 8a, object selection unit 9, and fusion false detection determination unit 10 are implemented by a processor 201, such as a CPU (Central Processing Unit), that executes a program stored in a memory 202. The program stored in the memory 202 may be, for example, a program 205 stored in a recording medium 204. In other words, this program can be said to cause a computer to execute the operation procedures of the components of the object detection device 100, 100a and the object detection method. The memory 202 is also used as a temporary storage device for each process executed by the processor 201. The memory 202 may be, for example, a non-volatile or volatile semiconductor memory such as RAM, ROM, flash memory, or EPROM, a magnetic disk, an optical disk, or a combination thereof. The processor 201 and the memory 202 are connected to a bus 203. The first detection device 1, the second detection device 2, and the vehicle information sensor 3 are connected to the bus 203 via a communication interface 206. The communication interface 206 connecting the first detection device 1, the second detection device 2, and the vehicle information sensor 3 to the bus 203 may be wired communication or wireless communication, and may consist of a single interface that transmits and receives multiple types of signals, or may consist of multiple interfaces with individual functions.

[0085] FIG. 11 is a schematic diagram showing another example of the hardware configuration of the object detection device 100, 100a according to the embodiment. In FIG. 11, the processing circuit 207 is connected to the bus 203, and the first detection device 1, the second detection device 2, and the vehicle information sensor 3 are connected to the bus 203 via a communication interface 206. When the processing circuit 207 is dedicated hardware, it may be, for example, a single circuit, a composite circuit, a programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof. Each of the functions of the object detection device 100, 100a may be realized by the processing circuit 207, or all of the functions may be realized by the processing circuit 207. Alternatively, some of the functions of the object detection device 100, 100a may be realized by dedicated hardware, and other parts may be realized by software or firmware.

[0086] Although the present application describes various exemplary embodiments, the various features, aspects, and functions described in one or more embodiments are not limited to application to a particular embodiment, but may be applied to the embodiments alone or in various combinations. Therefore, countless variations not illustrated are conceivable within the scope of the technology disclosed in this application, including, for example, cases where at least one component is modified, added, or omitted, and cases where at least one component is extracted and combined with a component of another embodiment. [Explanation of symbols]

[0087] 1 first detection device, 2 second detection device, 3 vehicle information sensor, 4 radar false detection determination unit, 5 radar filter processing unit, 6 camera filter processing unit, 7, 7a fusion processing unit, 8, 8a radar false detection removal unit, 9 object selection unit, 10 fusion false detection determination unit, 11 radar, 12 camera, 100, 100a object detection device, 101 vehicle, 102 radar detection range, 103 camera detection range, 104 fusion range, 105 noise, 106 noise, 201 processor, 202 memory, 203 bus, 204 recording medium, 205 program, 206 communication interface, 207 processing circuit.

Claims

1. An object detection device that detects an object based on information from a first detection device that outputs radar-detected object information including a radar-detected position, which is position information of the radar-detected object, for a radar-detected object that is an object detected based on an output of a radar, and information from a second detection device that outputs second-sensor-detected object information including a second-sensor-detected position, which is position information of the second-sensor-detected object, for a second-sensor-detected object that is an object detected based on an output of a second sensor different from a radar, a radar erroneous detection determination unit that determines whether the radar-detected object information acquired from the first detection device is suspected of being an erroneous detection, and sets a radar erroneous detection suspicion flag of the radar-detected object information that is determined to be suspected of being an erroneous detection to true; a fusion processing unit that outputs fusion object information including a fusion object position, which is position information of a fusion object, based on the second sensor-detected object information and the radar-detected object information acquired from the radar erroneous detection determination unit; and a radar erroneous detection deletion unit that deletes the fusion object information that is determined to be suspected of being an erroneous detection, The fusion processing unit includes: When it is determined that the radar-detected object and the second sensor-detected object are one fusion object, the radar-detected object information and the second sensor-detected object information are combined into one fusion object information, and a single sensor detection flag of the combined fusion object information is set to false; When it is determined that the fusion object is detected only by the radar, the radar-detected object information of the radar-detected object is set as the fusion object information, and the single-sensor detection flag of the fusion object information is set to true; The radar false detection elimination unit An object detection device, characterized in that the fusion object information for which the radar erroneous detection suspicion flag is true and the single sensor detection flag is true is deleted.

2. the radar erroneous detection determination unit determines that the radar-detected object information acquired from the first detection device is suspected to be an erroneous detection when the radar-detected object information acquired from the first detection device is determined to be at least one of a first erroneous detection and a second erroneous detection, When a change amount of the radar detected position of the radar-detected object toward the center in the left-right direction of the radar detection range is greater than a predetermined first change amount threshold over a predetermined first observation time, or, determining that the radar-detected object information includes information on a first lateral velocity that is a lateral movement velocity of the radar-detected object, and that the first erroneous detection has occurred when an absolute value of the first lateral velocity is greater than a predetermined first lateral velocity threshold value over the first observation time; When the radar-detected object information includes information on the size of the radar-detected object and information on the reflection intensity of the radar-detected object, and the size of the radar-detected object is larger than a predetermined size threshold and the reflection intensity of the radar-detected object is smaller than a predetermined reflection intensity threshold, or, When the radar-detected object information includes information on the reflection intensity of the radar-detected object, the distance from the radar to the radar detection position of the radar-detected object is smaller than a predetermined second distance threshold, and the reflection intensity of the radar-detected object is smaller than the predetermined reflection intensity threshold, or, 2. The object detection device according to claim 1, wherein the radar-detected object information includes information on the reliability of the radar-detected object and information on the reflection intensity of the radar-detected object, and the object detection device determines that the second erroneous detection has occurred when the reliability of the radar-detected object is smaller than a predetermined reliability threshold and the reflection intensity of the radar-detected object is smaller than the predetermined reflection intensity threshold.

3. The radar erroneous detection determination unit Acquire information on the vehicle's running speed, yaw rate, azimuth angle or running position from a vehicle information sensor as vehicle information, 3. The object detection device according to claim 2, wherein at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold is changed depending on the vehicle information.

4. The radar erroneous detection determination unit Acquire external environmental information including at least one of weather, time, temperature, road surface condition, and installation status of surrounding structures; 3. The object detection device according to claim 2, wherein at least one of the first observation time, the first change amount threshold, the first lateral velocity threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold is changed in accordance with the external environment information.

5. The radar erroneous detection determination unit Acquires map information and vehicle travel position information from the vehicle information sensor 3; 3. The object detection device according to claim 2, wherein at least one of the first observation time, the first change amount threshold, the first lateral speed threshold, the size threshold, the reflection intensity threshold, the second distance threshold, and the reliability threshold is changed in accordance with the map information and the traveling position information.

6. a fusion false detection determination unit that determines whether the fusion object information acquired from the fusion processing unit is suspected of being a false detection, and sets a fusion false detection suspicion flag of the fusion object information that is determined to be suspected of being a false detection to true; The fusion processing unit includes: a plurality of pieces of fusion object information in which the fusion object positions included in each piece of fusion object information are closer than a predetermined third distance threshold are combined into one piece of fusion object information, and output as new fusion object information; The fusion erroneous detection determination unit When the amount of change in the position of the fusion object toward the center in the left-right direction of the detection range of the radar is greater than a predetermined second change amount threshold over a predetermined second observation time, or, When the fusion object information includes information on a second lateral velocity, which is a lateral movement velocity of the fusion object, and the absolute value of the second lateral velocity is greater than a predetermined second lateral velocity threshold over the second observation time, the fusion object information is determined to be suspected of being a false detection; The radar false detection elimination unit Delete the fusion object information for which the radar false detection suspicion flag is true and the single sensor detection flag is true; 3. The object detection device according to claim 1, wherein the fusion object information for which the fusion erroneous detection suspicion flag is true and the single sensor detection flag is true is deleted.

7. The fusion erroneous detection determination unit Acquire information on the vehicle's running speed, yaw rate, azimuth angle or running position from a vehicle information sensor as vehicle information, The object detection device according to claim 6 , wherein at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold is changed in accordance with the vehicle information.

8. The fusion erroneous detection determination unit Acquire external environmental information including at least one of weather, time, temperature, road surface condition, and installation status of surrounding structures; The object detection device according to claim 6 , wherein at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold is changed in accordance with the external environment information.

9. The fusion erroneous detection determination unit Acquires map information and vehicle travel position information from the vehicle information sensor 3; 7. The object detection device according to claim 6, wherein at least one of the second observation time, the second change amount threshold, and the second lateral speed threshold is changed in accordance with the map information and the traveling position information.

10. An object detection method for detecting an object based on information from a first detection device that outputs radar-detected object information including a radar-detected position, which is position information of the radar-detected object, for a radar-detected object that is an object detected based on an output of a radar, and information from a second detection device that outputs second-sensor-detected object information including a second-sensor-detected position, which is position information of the second-sensor-detected object, for a second sensor-detected object that is an object detected based on an output of a second sensor different from a radar, a radar erroneous detection determination step of determining whether the radar-detected object information acquired from the first detection device is suspected of being an erroneous detection, and setting a radar erroneous detection suspicion flag of the radar-detected object information determined to be suspected of being an erroneous detection to true; a fusion processing step of outputting fusion object information including a fusion object position, which is position information of a fusion object, based on the second sensor detected object information and the radar detected object information processed in the radar erroneous detection determination step; a radar false detection deletion step of deleting the fusion object information determined to be suspected of being a false detection, The fusion processing step includes: When it is determined that the radar-detected object and the second sensor-detected object are one fusion object, the radar-detected object information and the second sensor-detected object information are combined into one fusion object information, and a single sensor detection flag of the combined fusion object information is set to false; When it is determined that the fusion object is detected only by the radar, the radar-detected object information of the radar-detected object is set as the fusion object information, and the single-sensor detection flag of the fusion object information is set to true; The radar false detection removal step includes: An object detection method, characterized in that the fusion object information for which the radar false detection suspicion flag is true and the single sensor detection flag is true is deleted.

11. An object detection program that causes a computer to execute the object detection method according to claim 10.

Images