Collision Avoidance Assist Device

Abstract

Reduce the processing load by appropriately selecting detection targets and appropriately execute collision avoidance assist control. The collision avoidance assist device includes: a front-side target information acquisition device that acquires front-side target information; a vehicle information acquisition device that acquires vehicle information including at least one of vehicle speed and yaw angular velocity or steering input value; and a control unit that, when a target satisfies a collision condition that holds when it is determined based on the front-side target information that there is a possibility of collision, executes at least one of alarm control and automatic braking control as collision avoidance assist control. The control unit selects a target that satisfies a specified selection condition from the targets included in the front-side target information, determines whether the collision condition holds for the selected target, and when making this determination, determines whether the host vehicle is turning based on the vehicle information, and changes the selection condition according to whether the host vehicle is not turning or is turning.

Description

Technical Field

[0001] The present invention relates to a collision avoidance assistance device capable of performing collision avoidance assistance control for avoiding a collision with a target existing in a range including at least the diagonally front and side of a vehicle, i.e., the front side range, or for reducing the impact caused by the collision. Background Art

[0002] Conventionally, a collision avoidance assistance device that performs collision avoidance assistance control when there is a target around a vehicle that the vehicle may collide with (hereinafter referred to as "collision possibility") is known. The collision avoidance assistance device includes a target information acquisition device that detects a target existing around the vehicle to obtain information related to the target (target information). The collision avoidance assistance device determines whether a collision condition (a condition that holds when there is a collision possibility) holds for each target included in the target information, and performs collision avoidance assistance control on the target for which the collision condition holds. According to this configuration, when the number of targets detected by the target information acquisition device increases, the amount of information in the target information increases, so the processing load involved in determining whether the collision condition holds increases. Therefore, a higher-performance computing device is required.

[0003] Therefore, for example, Patent Document 1 discloses a vehicle control device (hereinafter referred to as "existing device") that reduces the amount of information in target information. The target information acquisition device included in the existing device is composed of multiple sensors. Therefore, when detecting a target existing in a portion where the detection ranges of the multiple sensors overlap, due to deviations in ranging timing or ranging errors of each sensor, the positions (coordinates) of the targets included in the target information of each sensor are different. As a result, a situation where multiple positions are detected may occur. Each sensor is pre-assigned a reliability that reflects the detection accuracy of the target. The existing device is configured to: when the above situation occurs, adopt the target information of the sensor with the highest reliability (i.e., select the coordinates of the target included in the target information), and the target information of the remaining sensors is excluded, thereby reducing the amount of information in the target information. It should be noted that in Patent Document 1, the target information acquisition device and the target information are respectively described as "surrounding information sensor" and "surrounding information".

[0004] According to the existing device, the possibility of an unnecessary increase in the amount of information in the target information related to one target can be suppressed, but there is no research on reducing the amount of information in the target information related to multiple targets. Therefore, the existing device cannot solve the above problem (the problem that the stability of the collision avoidance assistance control may decrease when the number of detected targets is large).

[0005] Therefore, the collision avoidance assistance device is configured to: select a target that satisfies a specified selection condition when there are a large number of detected targets, and exclude the unselected targets from the target information, thereby reducing the amount of information in the target information.

[0006] However, when the target information acquisition device is configured to detect targets existing in the front side range of the vehicle, the uniformly set selection condition may not be able to appropriately select the targets with a collision possibility. That is, it is possible to select the targets with a lower collision possibility and not select the targets with a collision possibility. In this case, although the amount of information in the target information can be reduced, it may not be possible to appropriately perform the collision avoidance assistance control.

[0007] Prior art documents

[0008] Patent documents

[0009] Patent Document 1: Japanese Invention Patent Publication No. 6758438 Summary of the Invention

[0010] The present invention is an invention for solving the above problems. That is, one of the objects of the present invention is to provide a technology that can reduce the processing load by appropriately selecting detected targets and appropriately perform collision avoidance assistance control in a collision avoidance assistance device having a front side target information acquisition device capable of detecting targets existing in the front side range of the vehicle.

[0011] The collision avoidance assistance device of the present invention (hereinafter referred to as "the device of the present invention") includes: a forward target information acquisition device (11, 12) that detects a target existing in a prescribed forward range (Rf) including at least the front of the own vehicle, and acquires information related to the detected target as forward target information; a front side target information acquisition device (13L, 13R) that detects a target existing in a prescribed front side range (Rs, Rt) including at least the obliquely front and side of the own vehicle, and acquires information related to the detected target as front side target information; a vehicle information acquisition device (14, 15, 16) that acquires vehicle information including at least one of the speed of the own vehicle and the yaw angular velocity of the own vehicle or the steering input value of the own vehicle, the steering input value being an input value based on a steering operation; and a control unit (10) that, when a collision condition that is satisfied when the target satisfies a condition that it is determined that the own vehicle may collide with the target based on the forward target information and the front side target information is established, performs at least one of alarm control for issuing an alarm to the driver of the own vehicle and automatic braking control for automatically applying a braking force to the own vehicle as collision avoidance assistance control, wherein the control unit (10) is configured to: select a target that satisfies a prescribed selection condition from the targets included in the front side target information, and for each of the selected targets, determine whether the collision condition is established, and the control unit (10) is configured to: when making the determination, determine whether the own vehicle is turning based on the vehicle information, and change the selection condition according to a case where the own vehicle is not turning and a case where the own vehicle is turning.

[0012] Among the targets existing in the front side range, the targets having a collision possibility in the case of not turning and the targets having a collision possibility in the case of turning have different characteristics (for example, speed and position). Therefore, by changing the selection condition according to the case of not turning and the case of turning, it is easier to select the targets having a collision possibility. According to this configuration, the processing load can be reduced (that is, the number of targets to be determined whether the collision condition is satisfied is reduced), and the collision avoidance assistance control can be appropriately executed.

[0013] In one aspect of the present invention, the control unit (10) is configured to: when it is determined that the host vehicle is not turning, further determine whether the host vehicle is going straight. In a first case where the host vehicle is going straight, as the selection condition, it includes a condition that is satisfied when the target is a first target (40) having a speed within a specified first speed range. In a second case where the host vehicle is turning, as the selection condition, it includes a condition that is satisfied when the target is a second target (50, 60, 61) having a speed within a specified second speed range. The specified second speed range has a lower limit value less than the lower limit value of the first speed range and an upper limit value less than the upper limit value of the first speed range.

[0014] Typically, the targets that may cause a collision when the host vehicle is going straight (the first case) are the targets that approach the host vehicle at a relatively high speed from the blind spots on the left and right when entering an intersection. On the other hand, typically, the targets that may cause a collision when the host vehicle is turning (the second case) are the targets that are traversing or about to traverse the lane at a relatively low speed when turning left or right at an intersection. Here, the upper limit value and the lower limit value of the second speed range are less than the upper limit value and the lower limit value of the first speed range. Therefore, the first target is a target that moves at a speed higher than that of the second target. Thus, according to the above configuration, in both the first case and the second case, the targets that may cause a collision can be preferentially (appropriately) selected.

[0015] In one aspect of the present invention, the control unit (10) is configured to: in the first case, as the selection condition, it includes a condition that is satisfied when the first target (40) exists within a specified first specific range (Rs) in the front side range. In the second case, as the selection condition, it includes a condition that is satisfied when the second targets (50, 60, 61) exist within a specified second specific range (Rt) in the front side range that is narrower than the first specific range (Rs).

[0016] The first target moves at a speed higher than that of the second target. Therefore, by making the first specific range wider than the second specific range, in the case where the host vehicle is going straight (the first case), the possibility that the target that may cause a collision is not included in the first specific range can be reduced. On the other hand, the second target moves at a speed lower than that of the first target. Therefore, by making the second specific range narrower than the first specific range, the possibility that the target with a low collision possibility is included in the second specific range can be reduced. Thus, according to the above configuration, in both the first case and the second case, the targets that may cause a collision can be preferentially selected.

[0017] In one aspect of the present invention, the vehicle information acquisition device (16) acquires the steering input value, and the control unit (10) is configured to: in the second case, calculate the turning angle (θ) of the host vehicle from the start of turning to the current time point based on the steering input value, and when the turning angle (θ) exceeds a specified angle threshold (θth), set a range including a left range (Rtl) and a right range (Rtr) as the second specific range (Rt), where the left range (Rtl) at least includes the left front and the left side of the host vehicle, the right range (Rtr) at least includes the right front and the right side of the host vehicle, and when the turning angle (θ) is below the angle threshold (θth), set the range on the turning direction side of the host vehicle in the left range (Rtl) and the right range (Rtr) as the second specific range (Rt).

[0018] In the case where the turning angle exceeds the angle threshold, a target with a possibility of collision (i.e., a target that is crossing or about to cross the lane at a low speed when turning left or right at an intersection) is difficult to be detected by the front target information acquisition device and is located on either side of the front range. Therefore, by setting a range including both the left range and the right range as the second specific range, targets with a possibility of collision can be preferentially selected.

[0019] On the other hand, in the case where the turning angle is below the angle threshold, a "same-direction target moving in the same direction as the traveling direction of the host vehicle before starting to turn left or right" among the targets with a possibility of collision is difficult to be detected by the front information acquisition device and is located on the turning direction side with respect to the front range. However, a "opposite-direction target moving in the direction opposite to the above traveling direction" can be detected by the front information acquisition device and is located in the front range. Therefore, by setting the range on the turning direction side in the left range and the right range as the second specific range, a "target that cannot be detected by the front information acquisition device but has a possibility of collision" can be preferentially selected.

[0020] In one aspect of the present invention, the control unit (10) is configured to: when the number of targets satisfying the selection condition in the first case and the second case respectively exceeds a specified upper limit number (n), determine whether each target is continuously detected based on the front side target information, and calculate the reliability of each target based on the determination result, and as the selection condition, add a reliability condition that is established when the target is a high-reliability target having a reliability above a specified reliability threshold.

[0021] According to this configuration, targets with a higher accuracy of selection can be preferentially selected.

[0022] In one aspect of the present invention, the control unit (10) is configured to select objects among the objects included in the front side object information that are equal to or less than the upper limit number (n). The control unit (10) is configured to, in the first case, when the number of highly reliable objects exceeds the upper limit number (n), for each of the highly reliable objects, calculate a simple collision prediction time defined by the distance from the highly reliable object to the host vehicle and the speed of the highly reliable object itself, and select the number of highly reliable objects equal to the upper limit number (n) in the first priority order where the smaller the simple collision prediction time, the higher the priority. In the second case, when the number of highly reliable objects exceeds the upper limit number, select the number of highly reliable objects equal to the upper limit number (n) in the second priority order where the smaller the distance, the higher the priority.

[0023] When the host vehicle is going straight (the first case), even if an object is located at a relatively far position from the host vehicle, there is a possibility of collision if the speed of the object is high. Therefore, by selecting highly reliable objects based on an index (simple collision prediction time) that includes not only the distance from the host vehicle but also the speed of the object itself, it is possible to preferentially select objects with a possibility of collision.

[0024] On the other hand, when the host vehicle is turning (the second case), the speed of the object is less than that of the object in the first case. Therefore, by selecting highly reliable objects based on the distance from the host vehicle, it is possible to preferentially select objects with a possibility of collision.

[0025] In one aspect of the present invention, the control unit (10) is configured to select objects among the objects included in the front side object information that are equal to or less than the upper limit number (n). The control unit (10) is configured to, in the first case, when the number of highly reliable objects is less than the upper limit number (n), for each of the low-reliability objects that do not meet the reliability condition, calculate a simple collision prediction time defined by the distance from the low-reliability object to the host vehicle and the speed of the low-reliability object itself, and select the low-reliability objects in such a way that the sum of the number of highly reliable objects and the number of low-reliability objects is equal to the upper limit number in the first priority order where the smaller the simple collision prediction time, the higher the priority. In the second case, when the number of highly reliable objects is less than the upper limit number (n), select the low-reliability objects in such a way that the sum of the number of highly reliable objects and the number of low-reliability objects is equal to the upper limit number (n) in the second priority order where the smaller the distance, the higher the priority.

[0026] According to this configuration, even if the number of targets that satisfy the reliability condition (high-reliability targets) is less than the upper limit number, it is possible to further select targets from the targets that do not satisfy the reliability condition (low-reliability targets). Therefore, it is possible to avoid a situation where the number of selected targets is extremely small, and the configuration of "selecting targets equal to or less than the upper limit number" can be effectively utilized.

[0027] In one aspect of the present invention, the control unit (10) is configured to: in the second case, in addition to the condition that holds when the target is the second target, as the selection condition, it further includes a condition of selecting a specified number of the first targets (53, 62) that are present in a specified first specific range (Rs) in the front side range and are less than the upper limit number (n).

[0028] When the present vehicle is turning (second case), the present vehicle may also collide with a first target (typically, a target approaching the present vehicle at a relatively high speed). Therefore, according to the above configuration, it is also possible to select such a first target, and thus collision avoidance assist control can be more appropriately performed.

[0029] In the above description, in order to assist in understanding the invention, reference numerals used in the embodiments are added in parentheses to the constituent elements of the invention corresponding to the embodiments, but each constituent element of the invention is not limited to the embodiments defined by the reference numerals. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a schematic configuration diagram of a collision avoidance assist device according to an embodiment of the present invention.

[0031] Figure 2 is a diagram showing the detection ranges of a front radar sensor, a front camera sensor, and a front side radar sensor.

[0032] Figure 3 is a diagram showing a specific range set when the present vehicle is going straight.

[0033] Figure 4 is a diagram for explaining a method of selecting a target when the present vehicle is going straight.

[0034] Figure 5A is a diagram showing a specific range set when the turning angle of the present vehicle is small.

[0035] Figure 5B is a diagram showing a specific range set when the turning angle of the present vehicle is large.

[0036] Figure 6 is a diagram for explaining a method of selecting a target when the turning angle of the present vehicle is small.

[0037] Figure 7 This is a diagram for explaining the method of selecting a target when the turning amplitude of the vehicle is relatively large.

[0038] Figure 8 This is a flowchart showing the routine executed by the CPU of the collision avoidance assist ECU of the collision avoidance assist device.

[0039] Figure 9A This is a flowchart (one of them) showing the routine (target selection process) executed by the CPU.

[0040] Figure 9B This is a flowchart (the second one) showing the routine (target selection process) executed by the CPU.

[0041] Figure 10 This is a flowchart showing the routine (reliability process) executed by the CPU.

[0042] Figure 11 This is a flowchart showing the routine (priority process) executed by the CPU.

[0043] Description of Reference Numerals

[0044] 10: Collision Avoidance Assist ECU, 11: Front Radar Sensor, 12: Front Camera Sensor, 13L: Front Side Radar Sensor, 13R: Front Side Radar Sensor, 14: Vehicle Speed Sensor, 15: Yaw Rate Sensor, 16: Steering Angle Sensor, 20: Alarm ECU, 21: Buzzer, 30: Brake ECU, 31: Brake Actuator, 32: Friction Braking Mechanism. Detailed Embodiment

[0045] (Configuration)

[0046] Hereinafter, a collision avoidance assist device (hereinafter, also referred to as "this embodiment device") according to an embodiment of the present invention will be described with reference to the drawings. As Figure 1As shown in the figure, the present implementation device includes a collision avoidance assist ECU 10, an alarm ECU 20, and a braking ECU 30. The ECUs 10, 20, and 30 have a microcomputer as the main part and are connected in such a way that they can communicate with each other via a CAN (Controller Area Network) not shown in the figure. It should be noted that ECU is the abbreviation of Electronic Control Unit. The microcomputer includes a CPU (Central Processing Unit), a ROM (Read-Only Memory), a RAM (Random Access Memory), and interfaces, etc. The CPU realizes various functions by executing instructions (programs, routines) stored in the ROM. Several or all of the ECUs 10, 20, and 30 may also be integrated into one ECU as a controller. Hereinafter, the vehicle equipped with the present implementation device is referred to as "this vehicle".

[0047] The collision avoidance assist ECU 10 is connected to a front radar sensor 11, a front camera sensor 12, a front side radar sensor 13L, a front side radar sensor 13R, a vehicle speed sensor 14, a yaw rate sensor 15, and a steering angle sensor 16, and acquires signals output or generated by these sensors every predetermined cycle (50 [ms] in the present implementation). Hereinafter, the collision avoidance assist ECU 10 is also simply referred to as "ECU 10".

[0048] As Figure 2 shown, the front radar sensor 11 is provided at the center of the front bumper of this vehicle. The front radar sensor 11 has a function of detecting a target existing in the front range of this vehicle (strictly speaking, the range from the left front diagonal to the right front diagonal of this vehicle) and acquiring information related to the detected target. Specifically, the front radar sensor 11 irradiates millimeter-wave radio waves in the front direction of this vehicle, and when a target exists, receives the reflected wave from the target. The front radar sensor 11 calculates the presence or absence of a target and the relative relationship between this vehicle and the target based on the irradiation timing and reception timing of its radio waves, etc. The relative relationship between this vehicle and the target includes the distance from this vehicle to the target, the azimuth of the target relative to this vehicle, and the relative speed, etc. It should be noted that the target includes a moving target (a moving vehicle, a pedestrian, etc.) and a stationary target (a stationary vehicle, a pedestrian, a guardrail, a median strip, etc.).

[0049] Figure 2The range Rf represents the range (detection range) within which the front radar sensor 11 can detect a target. The range Rf is line-symmetric with respect to the front and rear axles of the host vehicle. The horizontal field of view angle of the front radar sensor 11 is, for example, approximately 100°, and the detection limit distance is, for example, approximately 120 [m]. It should be noted that in this specification, for ease of viewing the drawings, the scale of the host vehicle and the scales of the respective ranges have been changed for illustration.

[0050] The front camera sensor 12 is provided on the back of the inside rearview mirror (Room Mirror: inside rearview mirror / Rearview Mirror: rearview mirror) of the host vehicle. The front camera sensor 12 has a function of acquiring information related to a target existing in the front range of the host vehicle. Specifically, the front camera sensor 12 captures the scenery in front of the host vehicle and calculates the presence or absence of a target and the relative relationship between the host vehicle and the target based on the captured image data. The front camera sensor 12 is provided at a position slightly separated from the front radar sensor 11. However, in the present embodiment, the range within which the front camera sensor 12 can detect a target is set to be substantially the same as the range Rf.

[0051] The front radar sensor 11 detects a target existing within the range Rf and outputs information related to the target as "first front target information" to the ECU 10. The front radar sensor 11 is pre-set with an upper limit number of targets that can be output as the first front target information, for example, 17. When there are more targets than the upper limit number within the range Rf, the front radar sensor 11 detects the targets by a known method until the upper limit number is reached.

[0052] Similarly, the front camera sensor 12 detects a target existing within the range Rf and outputs information related to the target as "second front target information" to the ECU 10. The front camera sensor 12 is pre-set with an upper limit number of targets that can be output as the second front target information, for example, 17. When there are more targets than the upper limit number within the range Rf, the front camera sensor 12 detects the targets by a known method until the upper limit number is reached.

[0053] In this way, by using two sensors (the front radar sensor 11 and the front camera sensor 12) to detect a target existing within the range Rf, the detection accuracy of the target is improved (described later). Hereinafter, the front radar sensor 11 and the front camera sensor 12 may sometimes be collectively referred to as "front sensors 11 and 12".

[0054] The front side radar sensor 13L is provided at the left corner of the front bumper of the vehicle. The front side radar sensor 13L has a function of obtaining information related to a target present in the front side range on the left side of the vehicle (strictly speaking, the range from the front of the vehicle to the left rear obliquely, that is, the range including at least the left front obliquely and the left side of the vehicle).

[0055] The front side radar sensor 13R is provided at the right corner of the front bumper of the vehicle. The front side radar sensor 13R has a function of obtaining information related to a target present in the front side range on the right side of the vehicle (strictly speaking, the range from the front of the vehicle to the right rear obliquely, that is, the range including at least the right front obliquely and the right side of the vehicle).

[0056] The front side radar sensors 13L and 13R both calculate the presence or absence of a target and the relative relationship between the vehicle and the target in the same way as the front radar sensor 11. However, the front side radar sensors 13L and 13R are configured to detect only moving targets, which is different from the front radar sensor 11 in this regard. Specifically, the front side radar sensors 13L and 13R calculate the speed of the target (i.e., the ground speed) based on the relative relationship between the vehicle and the target, and detect only moving targets whose speed is above a specified speed threshold (e.g., 2 [km / h]) as targets.

[0057] Figure 2 The ranges Rl and Rr respectively represent the ranges in which the front side radar sensors 13L and 13R can detect targets. The range Rl and the range Rr are line-symmetric with respect to the longitudinal axis of the vehicle. The horizontal field of view angles of the front side radar sensors 13L and 13R are both, for example, about 150°, and the detection limit distances are both, for example, about 120 [m]. The ranges Rl and Rr are located on the side of the vehicle closer to the side than the range Rf. The range Rl and the range Rr partially overlap in front of the vehicle. In addition, the range Rl and the range Rf partially overlap in the range from the front of the vehicle to the left front obliquely, and the range Rr and the range Rf partially overlap in the range from the front of the vehicle to the right front obliquely.

[0058] The front side radar sensors 13L and 13R respectively detect targets (strictly speaking, moving targets) existing in ranges Rl and Rr, and output information related to the targets as "front side target information" to the ECU 10. The front side radar sensors 13L and 13R are preset with an upper limit on the total number of targets that can be output as front side target information, for example, 18. When there are more than the upper limit number of targets in range Rl or range Rr, the front side radar sensor 13L or 13R detects the targets by a known method until the upper limit number is reached. The front side radar sensors 13L and 13R are provided for the purpose of detecting targets with a possibility of collision that mainly exist outside the detection ranges Rf of the front sensors 11 and 12.

[0059] The vehicle speed sensor 14 generates a signal corresponding to the traveling speed (vehicle speed) of the own vehicle. The ECU 10 acquires the signal generated by the vehicle speed sensor 14 and calculates the vehicle speed based on this signal.

[0060] The yaw rate sensor 15 generates a signal corresponding to the yaw rate acting on the own vehicle. The ECU 10 acquires the signal generated by the yaw rate sensor 15 and calculates the yaw rate based on this signal. In the present embodiment, the yaw rate when the own vehicle is turning to the right direction is defined as positive, and the yaw rate when turning to the left direction is defined as negative.

[0061] The steering angle sensor 16 (steering input value acquisition device) detects the steering angle (steering input value) of the steering wheel of the own vehicle and outputs its detection signal to the ECU 10. In the present embodiment, the steering angle when the steering wheel is turning to the right direction is defined as positive, and the steering angle when turning to the left direction is defined as negative.

[0062] The alarm ECU 20 is connected to the buzzer 21. The buzzer 21 is built into an instrument panel (not shown). Hereinafter, the alarm ECU 20 will also be simply referred to as "ECU 20". The ECU 10 is configured to be able to send an alarm command to the ECU 20 (described later). When receiving the alarm command, the ECU 20 causes the buzzer 21 to sound according to this command. Thus, the ECU 10 can issue an alarm to the driver of the own vehicle via the ECU 20.

[0063] The brake ECU 30 is connected to the brake actuator 31. The brake actuator 31 is provided in a hydraulic circuit between a master cylinder (not shown) that pressurizes the working oil by the pedal force of the brake pedal and a friction brake mechanism 32 provided on each wheel. The friction brake mechanism 32 includes a brake disc 32a fixed to the wheel and a brake caliper 32b fixed to the vehicle body. The hydraulic pressure of the working oil supplied from the brake actuator 31 causes the wheel cylinder built into the brake caliper 32b to operate, thereby pressing the brake pad against the brake disc 32a to generate a friction braking force. Hereinafter, the brake ECU 30 is also referred to as "ECU30".

[0064] The ECU 10 is configured to send a braking command (described later) to the ECU 30. When receiving the braking command, the ECU 30 drives (controls) the brake actuator 31 according to the command. Thus, the ECU 10 can automatically apply braking force to the vehicle via the ECU 30.

[0065] (Work Details)

[0066] Next, the details of the operation of ECU10 are described. ECU10 obtains the first front object information from the front radar sensor 11, obtains the second front object information from the front camera sensor 12, and obtains the front side object information from the front side radar sensors 13L and 13R. ECU10 fuses these object information (Fusing). Specifically, based on multiple object information of objects repeatedly detected in at least two overlapping ranges of range Rf, range Rl and range Rr, one object information is generated. As a result, the detection accuracy of the repeatedly detected objects (that is, the accuracy of the object information) is improved. Afterwards, ECU10 determines whether the object satisfies the collision condition (a condition that is established when there is a possibility of collision) based on the fused object information, and performs collision avoidance auxiliary control when the collision condition is met. The collision condition will be described later.

[0067] Here, when the amount of information of the object mark information is large (i.e., the number of detected object marks is large), the load (computational load) of the subsequent processing (i.e., the fusion processing of the object mark information and the determination processing of whether the collision condition is established) will become large. Therefore, the ECU 10 is configured to: perform object mark selection processing on the object marks included in the front side object mark information, and exclude the object mark information of the object marks that are not selected in the processing from the front side object mark information, thereby reducing the load in the subsequent processing. The object mark selection processing is a processing of selecting the object mark in a manner that the object mark is less than the specified upper limit number n (4 in this embodiment). The processing is described in detail below. It should be noted that other processing of selecting the object mark can also be performed on the object mark included in the front object mark information, but the present embodiment is for the purpose of reducing the amount of information of the front side object mark information, so the description of this processing is omitted.

[0068] <Object Selection Process>

[0069] In the object selection process, it is desired to preferentially select objects with a high possibility of collision and exclude objects with a low possibility of collision. The characteristics of objects with a high possibility of collision vary depending on the driving state of the vehicle. Specifically, when the vehicle is going straight (excluding objects detected by the front sensors 11 and 12), typically, objects with a high possibility of collision are those that approach the vehicle at a high speed from the blind spots on the left and right when entering an intersection. In contrast, when the vehicle is turning (excluding objects detected by the front sensors 11 and 12), typically, objects with a high possibility of collision are those that are crossing or about to cross the lane at a low speed when turning left or right at an intersection. Here, the above-mentioned lane refers to the lane that the vehicle will enter after turning left or right.

[0070] (Selection Based on the Speed and Position of the Object)

[0071] Therefore, the ECU 10 changes the method of selecting objects according to whether the vehicle is going straight or turning. Specifically, when the vehicle is going straight, the ECU 10 selects "objects that exist in a relatively wide specific range Rs (refer to Figure 3 and Figure 4 . This will be described later.) and are moving at a high speed" among the objects included in the front-side object information. Hereinafter, objects moving at a high speed will also be simply referred to as "high-speed objects". On the other hand, when the vehicle is turning, the ECU 10 selects "objects that exist in a relatively narrow specific range Rt (refer to Figures 5A to 7 . This will be described later.) and are moving at a low speed". Hereinafter, objects moving at a low speed will also be simply referred to as "low-speed objects". In this embodiment, the speed (ground speed) range of the high-speed objects is set to include the legal speed range of the vehicle, and the speed range of the low-speed objects is set to include the average speed range of pedestrians. The lower limit value and the upper limit value of the speed range of the low-speed objects are respectively smaller than the lower limit value and the upper limit value of the speed range of the high-speed objects. It should be noted that the situation where the vehicle is going straight and the situation where the vehicle is turning are respectively examples of "the first situation" and "the second situation", the high-speed objects and their speed range are respectively examples of "the first object" and "the first speed range", and the low-speed objects and their speed range are respectively examples of "the second object" and "the second speed range".

[0072] Here, whether the vehicle is going straight or turning can be determined based on the turning radius of the vehicle at the current time point. That is, the ECU 10 obtains the vehicle speed and yaw rate at the current time point from the vehicle speed sensor 14 and the yaw rate sensor 15 respectively, and calculates the turning radius r by dividing the vehicle speed by the yaw rate. When the vehicle is turning to the right, the turning radius r is positive, and when it is turning to the left, the turning radius r is negative. When the magnitude of the turning radius r is less than or equal to a specified first radius threshold r1th, the ECU 10 determines that the vehicle is turning. When the magnitude of the turning radius r exceeds a specified second radius threshold r2th (> r1th), the ECU 10 determines that the vehicle is going straight. On the other hand, when r1th < |r| ≤ r2th holds, the ECU 10 determines that it is impossible to determine whether the vehicle is going straight or turning, and proceeds with the subsequent processing without performing the target selection process. Here, the case where r1th < |r| ≤ r2th holds means, for example, that the steering wheel of the vehicle is temporarily steered due to an accidental steering operation. It should be noted that the vehicle speed and yaw rate are examples of "vehicle information", and the vehicle speed sensor 14 and the yaw rate sensor 15 are examples of "vehicle information acquisition devices".

[0073] It should be noted that, regarding whether the vehicle is going straight or turning, it is not limited to the configuration determined based on the turning radius, and it can also be configured to be determined based on the vehicle speed, steering angle, and steering angular velocity at the current time point. The steering angular velocity can be obtained by a steering angular velocity sensor (not shown). The steering angle and steering angular velocity are examples of the "steering input value" of the vehicle, and the "steering input value" is an input value based on a steering operation (operation of the steering wheel). In this case, the vehicle speed, steering angle, and steering angular velocity are examples of "vehicle information", and the vehicle speed sensor 14, the steering angle sensor 16, and the steering angular velocity sensor are examples of "vehicle information acquisition devices".

[0074] As Figure 3 shown, the specific range Rs is a range provided within the ranges Rl and Rr, and its size and shape can be determined by obtaining the distribution of targets that may collide when the vehicle V is going straight based on experiments or simulations. The high-speed target has a long moving distance per unit time. Therefore, the specific range Rs is set such that the length in the radial direction is equal to the detection limit distance of the front side radar sensors 13L and 13R, and mainly includes the obliquely front and side of the vehicle V. The point P1 of the specific range Rs is located on the extension line of the front and rear axles of the vehicle V. The length from the center of the front end of the vehicle V to the point P1 can be set to about 10 [m] based on the lane widths of multiple lanes intersecting the driving lane of the vehicle V, for example.

[0075] Figure 4 This shows an example of a method for selecting a target when the vehicle V is going straight. As Figure 4 shown, there are four targets 40 to 43 within the ranges Rl and Rr (that is, the target information of targets 40 to 43 is included in the front side target information). Targets 40 and 42 are high-speed targets, and targets 41 and 43 are low-speed targets.

[0076] The ECU 10 selects target 40, which is a high-speed target existing in the specific range Rs, from these targets 40 to 43. Targets 41 and 42 among the unselected targets 41 to 43 also exist within the range Rf (the detection range of the front sensors 11 and 12), so they are detected by the front sensors 11 and 12. In this way, by setting the specific range Rs not to include "the range that can be appropriately detected by the front sensors 11 and 12", it is possible to preferentially select "targets that cannot be detected by the front sensors 11 and 12 but have a possibility of collision". On the other hand, although target 43 exists in the specific range Rs, its speed is low and the moving distance per unit time is very short. Therefore, even if it moves in the direction approaching the vehicle V, the possibility of collision is extremely low. Therefore, by not selecting the low-speed targets existing in the specific range Rs, it is possible to exclude targets with a low possibility of collision.

[0077] On the other hand, as Figure 5A and Figure 5B shown, the specific range Rt is a range set within the range Rl and / or Rr, and its size and shape can be determined by obtaining the distribution of targets with a possibility of collision when the vehicle V is turning based on experiments or simulations. The moving distance per unit time of the low-speed targets is short. Therefore, the specific range Rt is narrower than the specific range Rs and is set to be included within the specific range Rs in this embodiment. Depending on the turning amplitude of the vehicle V, the specific range Rt sometimes includes only one of the left range Rtl or the right range Rtr, and sometimes includes both the left range Rtl and the right range Rtr.

[0078] Figure 5A shows a situation where the vehicle V is turning to the right with a small turning amplitude, Figure 5B shows a situation where the vehicle V is turning to the right with a large turning amplitude. As Figure 5A and Figure 5BAs shown, when the turning amplitude is small, the ECU 10 sets the range on the turning direction side (in this example, the right range Rtr) in the left range Rl and the right range Rr as the specific range Rt. When the turning amplitude is large, the ECU 10 sets both the left range Rtl and the right range Rtr as the specific range Rt. Hereinafter, sometimes the former specific range Rt is referred to as the "turning direction specific range Rt" and the latter specific range Rt is referred to as the "two-direction specific range Rt". The size of the turning amplitude can be determined based on whether the turning angle θ (the angle turned from the time point when the own vehicle V starts turning to the current time point) exceeds a specified angle threshold θth. The turning angle θ can be calculated by dividing the moving distance from the time point when the own vehicle V starts turning to the current time point by the turning radius r. The angle threshold θth can be determined in advance based on the horizontal field of view angles of the front sensors 11 and 12. It should be noted that a turning angle correlation value can also be used instead of the turning angle θ. In this case, the size of the turning amplitude can be determined based on whether the turning angle correlation value exceeds a specified correlation value threshold. The turning angle correlation value can be calculated by performing a time integration on the product of the vehicle speed and the steering angle (the angle obtained from the steering angle sensor 16).

[0079] The left range Rtl is located within the range Rl, and the right range Rtr is located within the range Rr. The left range Rtl and the right range Rtr are line-symmetric with respect to the front and rear axles of the own vehicle V. The vehicle width direction length of the range including the left range Rtl and the right range Rtr is constant in the front and rear axle direction, for example, about 25 [m], and the maximum value of the front and rear axial length is, for example, about 20 [m]. The length from the center of the front end of the own vehicle V to the point P2 can be set to about 5 [m], for example.

[0080] Figure 6 An example of the method for selecting an object in the case of a small turning amplitude (θ ≤ θth) in the rightward direction is shown. As Figure 6 shown, there are four objects 50 to 53 within the ranges Rl and Rr (that is, the object information including the objects 50 to 53 is included in the front side object information). The objects 50 to 52 are low-speed objects, and the object 53 is a high-speed object.

[0081] The ECU 10 selects the low-speed object, that is, the object 50, which exists in the turning direction specific range Rt (in this example, the right range Rtr) among these objects 50 to 53. Among the unselected objects 51 to 53, the object 51 also exists within the range Rf, and thus is detected by the front sensors 11 and 12. In this way, by setting the specific range Rt to not include the "range that can be appropriately detected by the front sensors 11 and 12", it is possible to preferentially select the "objects that cannot be detected by the front sensors 11 and 12 but have a collision possibility".

[0082] More specifically, the object 50 is a same-direction low-speed object moving in the same direction as the traveling direction of the vehicle V before starting a right turn, and the object 51 is an oncoming-direction low-speed object moving in the direction opposite to the above-mentioned traveling direction. These objects 50 and 51 are objects that are crossing the lane (the lane that the vehicle V will enter after the right turn) or about to cross, so there is a possibility of collision for both. Generally speaking, in the case of a small turning angle, the oncoming-direction low-speed object (object 51 in this example) is detected by the front sensors 11 and 12, so the specific turning direction range Rt is set to be able to mainly detect the shape and size of the same-direction low-speed object (object 50 in this example).

[0083] On the other hand, although the object 52 exists in the left range Rtl (that is, the range on the side opposite to the turning direction), its speed is low and the moving distance per unit time is very short. Therefore, even if it moves in the turning direction of the vehicle V, the possibility of collision is extremely low. Therefore, by not selecting the low-speed objects in the left range Rtl and the right range Rtr on the side opposite to the turning direction, objects with a low possibility of collision can be excluded.

[0084] In contrast, the object 53 is a high-speed object that also exists in the specific range Rs. Compared with the objects 50 and 51, the possibility of collision is low, but if it continues its current moving state, it may collide with the vehicle V (side collision). Therefore, the ECU 10 is configured to: when the number of low-speed objects existing in the specific range Rt is less than n, be able to additionally select the high-speed objects existing in the specific range Rs (described later). It should be noted that the object 53 is typically a vehicle traveling in the lane that the vehicle V enters after a right turn in a direction approaching the vehicle V.

[0085] Figure 7 An example of the method for selecting objects in the case of a large turning angle (θ > θth) indicating the right direction is shown. As Figure 7 shown, there are three objects 60 to 62 in the ranges Rl and Rr (that is, the object information including the objects 60 to 62 is included in the front-side object information). The objects 60 and 61 are low-speed objects, and the object 62 is a high-speed object.

[0086] The ECU 10 selects the low-speed objects 60 and 61 existing in the two-direction specific range Rt among these objects 60 to 62. More specifically, the object 60 is a same-direction low-speed object, and the object 61 is an oncoming-direction low-speed object, both having a possibility of collision. Generally speaking, as the turning angle becomes larger, the oncoming-direction low-speed object (object 61 in this example) becomes difficult to be detected by the front sensors 11 and 12. Therefore, in the case of a large turning angle, the two-direction specific range Rt is set to be able to detect not only the same-direction low-speed object (object 60 in this example) but also the shape and size of the oncoming-direction low-speed object.

[0087] In contrast, the object 62 is a high-speed object that also exists within a specific range Rs. Compared with the objects 60 and 61, the possibility of collision is low. However, if it continues its current moving state, it may collide with the present vehicle V (side collision). Therefore, the ECU 10 is configured in the same way as in the case of a smaller turning amplitude: when the number of low-speed objects within a specific range Rt is less than n, it can additionally select high-speed objects existing within the specific range Rs (described later). It should be noted that the object 62 is typically a vehicle traveling in the oncoming lane (the lane opposite to the lane in which the present vehicle V was traveling before starting to turn right) in a direction approaching the present vehicle V.

[0088] When the present vehicle V is going straight, the ECU 10 counts the high-speed objects among the objects within the specific range Rs. When the number of high-speed objects is n (= 4) or less, the ECU 10 selects all of these high-speed objects. Then, the object information of the objects that have not been selected from the objects existing within the ranges Rl and Rr is excluded (deleted) from the front-side object information. Hereinafter, the unselected objects will also be referred to as "non-selected objects". In Figure 4 the example, the only high-speed object within the specific range Rs is the object 40 (i.e., 4 or less), so the ECU 10 selects the object 40 and excludes the object information of the non-selected objects 41 to 43 from the front-side object information. Hereinafter, "excluding the object information of non-selected objects from the front-side object information" will also be simply referred to as "excluding non-selected objects".

[0089] When the present vehicle V is turning (regardless of the turning amplitude), the ECU 10 counts the low-speed objects among the objects within the specific range Rt. When the number of low-speed objects is equal to n, all of these low-speed objects are selected and non-selected objects are excluded. On the other hand, when the number of low-speed objects within the specific range Rt is less than n in the same situation, the ECU 10 first selects all of these low-speed objects. Then, if there are high-speed objects within the specific range Rs, high-speed objects are additionally selected until the total number of selected objects reaches n, and then non-selected objects are excluded. In Figure 6 the example, the only low-speed object within the specific range Rt is the object 50 (i.e., less than 4), so the ECU 10 first selects the object 50. Then, since there is a high-speed object 53 within the specific range Rs, the object 53 is additionally selected, and then the non-selected objects 51 and 52 are excluded. The additional selection method for high-speed objects will be described later.

[0090] As described above, when the number of high-speed targets within a specific range Rs is less than n while the vehicle V is going straight, even if there are low-speed targets within the specific range Rs, the ECU 10 will not perform the process of selecting such low-speed targets in an additional manner. This is because when the vehicle V is going straight, the possibility of the vehicle V colliding with the low-speed targets within the specific range Rs is extremely low (refer to Figure 4 target 43).

[0091] (Reliability processing)

[0092] In contrast, when the number of high-speed targets within the specific range Rs exceeds n while the vehicle is going straight, the ECU 10 calculates the reliability for each of these high-speed targets and performs a reliability process of extracting the targets with a reliability above a specified reliability threshold. Hereinafter, the method of calculating the reliability will be specifically described.

[0093] Reliability is an index indicating the accuracy of a target and can take values from 0 to 200 points (Point). The initial value of the reliability of a target (i.e., the reliability in the cycle when first detected by the front side radar sensors 13L and 13R) is set to 30 points. When a target is detected in consecutive cycles, 60 points are added to each cycle. However, when the reliability reaches 200 points, it will not be further increased. On the other hand, when a target is no longer detected midway, 10 points are subtracted when it is not detected for one cycle, 30 points are subtracted when it is not detected for two cycles, and 100 points are subtracted for each cycle when it is not detected for three or more cycles. However, when the reliability reaches 0 points, it will not be further decreased.

[0094] For example, when a certain target is further detected in two consecutive cycles after the cycle when first detected, the reliability of this target is 150 points (30 + 60×2), and when it is further detected in two consecutive cycles thereafter, the reliability of this target is 200 points. On the other hand, when another target is not detected for two cycles at the time point when the reliability is 200 points, the reliability of this target is 170 points (200 - 30), and when it is not detected for four cycles at the above time point, the reliability of this target is 0 points (200 - 200). It should be noted that the reliability threshold can be set to any value greater than the initial value of the reliability (30 points) based on experiments or simulations.

[0095] The ECU 10 calculates the reliability of each target as described above and extracts the targets with a reliability above the reliability threshold (hereinafter, also referred to as "high-reliability targets"). It should be noted that hereinafter, the targets with a reliability less than the reliability threshold are also referred to as "low-reliability targets".

[0096] After the reliability process is completed, when the number of highly reliable targets extracted is equal to n, ECU10 selects all these highly reliable targets and excludes non-selected targets. For example, consider the following situation: there are 9 high-speed targets and 3 low-speed targets within the ranges Rl and Rr, and 7 of the high-speed targets are within a specific range Rs. In this case, the number of highly reliable targets extracted through the reliability process is 4 (=n). In this situation, ECU10 selects these 4 highly reliable targets, and then excludes the remaining 8 non-selected targets (3 low-reliability targets, 2 high-speed targets that have not undergone the reliability process, and 3 low-speed targets).

[0097] In contrast, when the number of highly reliable targets extracted is not n (i.e., the number of highly reliable targets is less than or greater than n), ECU10 performs a priority process (described later) to make the number of selected targets equal to n.

[0098] On the other hand, when the number of low-speed targets within a specific range Rt exceeds n while the vehicle is turning, ECU10 performs the above-mentioned reliability process for each of these low-speed targets.

[0099] After the reliability process is completed, when the number of highly reliable targets extracted is equal to n, ECU10 selects all these highly reliable targets and excludes non-selected targets. For example, consider the following situation: there are 11 low-speed targets and 5 high-speed targets within the ranges Rl and Rr, and 9 of the low-speed targets are within a specific range Rt. In this case, the number of highly reliable targets extracted through the reliability process is 4 (=n). In this situation, ECU10 selects these 4 highly reliable targets, and then excludes the remaining 12 non-selected targets (5 low-reliability targets, 2 low-speed targets that have not undergone the reliability process, and 5 high-speed targets).

[0100] In contrast, when the number of highly reliable targets extracted is not n (i.e., the number of highly reliable targets is less than or greater than n), ECU10 performs a priority process (described later) to make the number of selected targets equal to n.

[0101] (Priority process)

[0102] As described above, when the number of high-speed targets within a specific range Rs exceeds n (when going straight) or the number of low-speed targets within a specific range Rt exceeds n (when turning), the priority process is implemented when the number of highly reliable targets extracted through the reliability process is not n. The following is an explanation in order.

[0103] When the number of highly reliable objects is less than n while the vehicle is going straight, the ECU 10 selects all these highly reliable objects. Then, the simple collision prediction time is calculated separately for the remaining low-reliability objects. The simple collision prediction time is a simplified version of the collision prediction time (the time predicted until the vehicle collides with the object). The prediction accuracy of the simple collision prediction time is not as accurate as that of the collision prediction time, but it can calculate the urgency of the vehicle colliding with the object with a lower load. Hereinafter, the collision prediction time is referred to as "TTC" (Time To Collision), and the simple collision prediction time is referred to as "simple TTC".

[0104] The simple TTC can be calculated by dividing the "distance from the vehicle to the object" by the "velocity component of the object approaching the vehicle" based on the information of the front-side objects. The ECU 10 selects low-reliability objects in the order of the first priority where the smaller the simple TTC, the higher the priority, until the total number of selected objects reaches n. That is, the ECU 10 selects low-reliability objects in such a way that the sum of the number of selected highly reliable objects and the number of low-reliability objects selected in the order of the first priority is n. The above is the priority processing when the number of highly reliable objects is less than n while the vehicle is going straight.

[0105] After the priority processing is completed, the ECU 10 excludes non-selected objects. For example, consider the following situation: there are 9 high-speed objects and 3 low-speed objects within the ranges Rl and Rr, and 7 of the high-speed objects are within a specific range Rs. In this case, the number of highly reliable objects extracted through the reliability processing is 1 (<n). In this case, the ECU 10 selects this 1 highly reliable object, and selects 3 low-reliability objects from the 6 low-reliability objects within the specific range Rs in ascending order of the simple TTC (in the order of the first priority). Thus, the total number of selected objects is 4. After that, the ECU 10 excludes the remaining 8 non-selected objects (3 low-reliability objects, 2 high-speed objects that have not undergone reliability processing, and 3 low-speed objects).

[0106] On the other hand, when the number of highly reliable targets exceeds n while the vehicle is going straight, the ECU 10 calculates the simple TTC for these highly reliable targets respectively, and performs priority processing to select n highly reliable targets in the order of the first priority. After the priority processing is completed, the ECU 10 excludes the unselected targets. For example, consider the following situation: when there are 9 high-speed targets and 3 low-speed targets within the ranges Rl and Rr, and 7 of the high-speed targets are within the specific range Rs, the number of highly reliable targets extracted by the reliability processing is 5 (>n). In this case, the ECU 10 selects 4 highly reliable targets from these 5 highly reliable targets in the order of the first priority. After that, the ECU 10 excludes the remaining 8 unselected targets (1 highly reliable target, 2 low-reliability targets, 2 high-speed targets that have not undergone reliability processing, and 3 low-speed targets).

[0107] In contrast, when the number of highly reliable targets is less than n while the vehicle is turning, the ECU 10 selects all of these highly reliable targets. Then, the remaining low-reliability targets are selected in the order of the second priority where the smaller the distance from the vehicle, the higher the priority, until the total number of selected targets reaches n.

[0108] Here, when the distances from the vehicle to multiple low-reliability targets are the same, the number of selected targets may not be n (hereinafter, multiple targets with the same distance from the vehicle are also referred to as "same-distance targets"). The low-speed targets include not only pedestrians but also vehicles traveling at low speeds. However, when the vehicle is turning, it is desired to perform collision avoidance assist control for pedestrians prior to vehicles. Therefore, the ECU 10 is configured to: when the number of selected targets is not n even if the low-reliability targets are selected in the order of the second priority, identify pedestrians from the same-distance targets, and select pedestrians prior to vehicles to make the number of selected targets n. The following is a specific description.

[0109] First, the ECU 10 determines whether the micro-Doppler condition is satisfied for each of the same-distance targets. The micro-Doppler condition is a condition that holds when a pedestrian can be identified by micro-Doppler determination. Micro-Doppler determination is a well-known determination for identifying pedestrians and is performed using microwaves. That is, the front-side radar sensors 13L and 13R are configured to be able to irradiate microwaves in addition to millimeter-waveband radio waves. When the micro-Doppler condition is satisfied for a certain target, the ECU 10 preferentially selects that target (the target identified as a pedestrian).

[0110] When the number of selected targets is not n because there are multiple pedestrians or no pedestrians among the targets at the same distance, the ECU 10 determines whether the speed condition is satisfied for each of the targets at the same distance. The speed condition is a condition satisfied by the target having the speed closest to the reference speed (for example, 5 [km / h] in the present embodiment). The ECU 10 preferentially selects the targets for which the speed condition is satisfied.

[0111] When the number of selected targets is not n because there are multiple targets satisfying the speed condition among the targets at the same distance, the ECU 10 determines whether the size condition is satisfied for each of the targets at the same distance. The size condition is a condition satisfied when each of the longitudinal, lateral, and height dimensions of the target is 1 m or less. When the size condition is satisfied for a certain target, the ECU 10 preferentially selects that target. When the number of selected targets is still not n even after the above selection, the ECU 10 excludes the targets that satisfy the size condition, the targets that satisfy the speed condition, and the targets that satisfy the micro-Doppler condition in this order until the number of selected targets becomes n or less. The above is the priority processing when the number of highly reliable targets is less than n while the vehicle is turning.

[0112] After the priority processing is completed, the ECU 10 excludes the non-selected targets. For example, consider the following situation: there are 11 low-speed targets and 5 high-speed targets within the ranges Rl and Rr, and 9 of the low-speed targets are within the specific range Rt. In this case, the number of highly reliable targets extracted through the reliability processing is 1 (<n). In this case, the ECU 10 selects the 1 highly reliable target and selects 3 low-reliable targets from the 8 low-reliable targets within the specific range Rt in ascending order of distance (in the second priority order). Thus, the total number of selected targets is 4. It should be noted that when it is impossible to select 3 low-reliable targets even by selecting in the second priority order because there are multiple targets at the same distance within the specific range Rt, the ECU 10 attempts to apply the micro-Doppler condition, the speed condition, and the size condition in sequence to select 3 low-reliable targets. After that, the remaining 12 non-selected targets (5 low-reliable targets, 2 low-speed targets that have not undergone reliability processing, and 5 high-speed targets) are excluded.

[0113] On the other hand, when the number of highly reliable targets exceeds n while the vehicle is turning, the ECU 10 selects n highly reliable targets in the second priority order. When the number of selected targets is still not n even when selecting the highly reliable targets in the second priority order, the ECU 10 attempts to apply the above micro-Doppler condition, speed condition, and size condition in sequence to select the highly reliable targets so that the number of selected targets becomes n. The above is the priority processing when the number of highly reliable targets exceeds n while the vehicle is turning.

[0114] After the priority processing is completed, the ECU 10 excludes non-selected targets. For example, consider the following situation: when there are 11 low-speed targets and 5 high-speed targets within the ranges Rl and Rr, and 9 of the low-speed targets are within a specific range Rt, the number of high-reliability targets extracted through reliability processing is 5 (>n). In this case, the ECU 10 selects 4 high-reliability targets from these 5 high-reliability targets in the order of the second priority. The processing for the case where the number of selected high-reliability targets is not 4 is as described above. After that, the ECU 10 excludes the remaining 12 non-selected targets (1 high-reliability target, 4 low-reliability targets, 2 low-speed targets that have not undergone reliability processing, and 5 high-speed targets).

[0115] (Additional selection processing for high-speed targets during turning)

[0116] Although the targets with a collision possibility when the vehicle is turning are mainly low-speed targets within a specific range Rt, as described above, high-speed targets within a specific range Rs (although the collision possibility is lower than that of low-speed targets) also have a collision possibility. Therefore, when the number of low-speed targets within the specific range Rt is less than n and there are high-speed targets within the specific range Rs, the ECU 10 is configured to be able to select such high-speed targets in an additional manner until the number of selected targets reaches n.

[0117] Specifically, in the above case, for each of the high-speed targets within the specific range Rs, the ECU 10 performs reliability processing to extract high-speed targets with a reliability above the reliability threshold. When the number of extracted high-reliability targets is equal to the number nr obtained by subtracting the number of low-speed targets within the specific range Rt from n, the ECU 10 selects all these nr high-reliability targets and excludes non-selected targets. For example, when the number of low-speed targets within the specific range Rt is 1, the number of high-speed targets within the specific range Rs is 5, and the number of high-reliability targets among these 5 high-speed targets is 3 (=nr), the ECU 10 selects these 3 high-reliability targets. Thus, the total number of selected targets is 4. After that, the ECU 10 excludes the remaining 2 non-selected targets (2 low-reliability targets).

[0118] On the other hand, when the number of highly reliable targets extracted is less than nr, ECU10 selects all of these highly reliable targets. Then, when there are low-reliability targets (high-speed targets not extracted through reliability processing), ECU10 calculates the simple TTC for each of these low-reliability targets and selects them in the order of the first priority so that the total number of selected targets does not exceed n. After that, ECU10 excludes the non-selected targets. For example, when the number of low-speed targets within a specific range Rt is 1, the number of high-speed targets within a specific range Rs is 5, and the number of highly reliable targets among these 5 high-speed targets is 1 (<nr = 3), ECU10 selects this 1 highly reliable target. Then, 2 low-reliability targets are selected from the 4 low-reliability targets in the order of the first priority. Thus, the total number of selected targets is 4. After that, ECU10 excludes the remaining 2 non-selected targets (2 low-reliability targets).

[0119] On the other hand, when there are no low-reliability targets when the number of highly reliable targets extracted is less than nr, ECU10 selects all of these highly reliable targets, and then excludes the non-selected targets (for example, targets existing outside the specific range Rt).

[0120] In contrast, when the number of highly reliable targets extracted exceeds nr, ECU10 calculates the simple TTC for each of these highly reliable targets and selects n highly reliable targets in the order of the first priority. After that, ECU10 excludes the non-selected targets. For example, when the number of low-speed targets within a specific range Rt is 1, the number of high-speed targets within a specific range Rs is 5, and the number of highly reliable targets among these 5 high-speed targets is 4 (>nr = 3), ECU10 selects 3 highly reliable targets from these 4 highly reliable targets in the order of the first priority. Thus, the total number of selected targets is 4. After that, ECU10 excludes the remaining 2 non-selected targets (1 highly reliable target and 1 low-reliability target).

[0121] The above is the description of the target selection process. The first front target information, the second front target information, and the front side target information with reduced information volume through the target selection process are fused through the fusion process. After that, it is determined for each of the fused targets whether the collision condition can be established.

[0122] The collision conditions include an alarm condition and an automatic braking condition. The alarm condition is a condition that holds when a target exists within a specified alarm range (i.e., there is a possibility of colliding with the target) while the host vehicle is going straight. The alarm range can be set to the same shape and size as a specific range Rs, or can be set to any range within the specific range Rs. When a fused target exists within the alarm range while the host vehicle is going straight, the alarm condition holds for that target. In this case, the ECU 10 executes alarm control to issue an alarm to the driver as collision avoidance assist control. Specifically, when the alarm condition holds, the ECU 10 sends an alarm command to the ECU 20. When the ECU 20 receives the alarm command, it makes the buzzer 21 sound to issue an alarm to the driver, thereby executing the alarm control.

[0123] On the other hand, the automatic braking condition is a condition that holds when the time to collision (TTC) with a target is below a specified TTC threshold (i.e., there is a possibility of colliding with the target). Hereinafter, the calculation method of the TTC will be specifically described. First, the ECU 10 calculates the trajectory of the host vehicle and the trajectory of the target respectively. The trajectory of the host vehicle can be calculated based on the turning radius r of the host vehicle. The trajectory of the target can be calculated based on the movement of "the position of the target included in the fused target information". The ECU 10 determines whether the host vehicle will collide with the target on the condition that the host vehicle maintains its current driving state and the target maintains its current moving state. When it is determined that a collision will occur, the ECU 10 calculates the TTC for that target. This TTC can be calculated by dividing the distance from the host vehicle to "the point where it is determined that a collision with the target will occur" by the vehicle speed.

[0124] When the TTC for a fused target is below the TTC threshold, the automatic braking condition holds for that target. In this case, the ECU 10 executes automatic braking control to automatically apply a braking force to the host vehicle as collision avoidance assist control. Specifically, when the automatic braking condition holds, the ECU 10 calculates the target deceleration required to stop the vehicle in front of a specified distance from the target, and sends a braking command including the target deceleration to the ECU 30. When receiving the braking command, the ECU 30 controls the brake actuator 31 so that the actual acceleration matches the target deceleration, causing frictional braking forces to be generated on each wheel, thereby executing the automatic braking control.

[0125] Thus, for collision conditions (especially automatic braking conditions), the amount of processing for determining whether they are satisfied is large. Therefore, if the number of targets to be determined is large, the processing load will increase. However, according to the present embodiment device, the amount of information of the front-side target information is reduced by the target selection process, so the processing load when determining the collision conditions is reduced. In addition, in the target selection process, the method of selecting targets is changed according to whether the host vehicle is going straight or turning. Therefore, in either case, targets with a possibility of collision can be appropriately selected (in other words, the possibility of excluding targets with a possibility of collision or selecting targets with a low possibility of collision can be significantly reduced). Therefore, according to this configuration, the amount of information of the front-side target information can be reduced and the collision avoidance assist control can be appropriately executed.

[0126] (Specific operations)

[0127] Next, the specific operations of the CPU of the ECU 10 will be described. During the period when the ignition switch is in the ON position, the CPU executes the Figures 8 to 11 routine shown in the flowchart at every predetermined time.

[0128] When the predetermined timing arrives, the CPU starts processing from step 800 of Figure 8 and proceeds to step 805 and step 810. In step 805, the CPU acquires the first front target information from the front radar sensor 11 and acquires the second front target information from the front camera sensor 12. After that, the CPU proceeds to step 820 (described later). On the other hand, in step 810, the CPU acquires the front-side target information from the front-side radar sensors 13L and 13R.

[0129] After step 810 ends, the CPU proceeds to step 815 to perform the target selection process. That is, the CPU starts processing from step 900 (refer to Figure 9A ) and proceeds to step 905. In step 905, the CPU determines whether the host vehicle is turning based on the turning radius r. When not turning (|r| > r1th), the CPU determines "No" in step 905 and proceeds to step 910.

[0130] In step 910, the CPU determines whether the host vehicle is going straight based on the turning radius r. When not going straight (|r| ≤ r2th), the CPU determines "No" in step 910 (that is, it is determined that it is impossible to determine whether the host vehicle is going straight or turning), and proceeds to step 820 via step 995 (refer to Figure 8)(Described later). In this case, the target selection process is not performed. On the other hand, when going straight (|r| > r2th), the CPU determines "Yes" in step 910 and proceeds to step 915.

[0131] In step 915, the CPU determines whether the number of high-speed targets within a specific range Rs is equal to or less than the upper limit number n. Here, the "high-speed targets within the specific range Rs" refers to, in other words, targets that satisfy both the "conditions that hold when the target is a high-speed target" and the "conditions that hold when the high-speed target exists within the specific range Rs". When the number of high-speed targets within the specific range Rs is equal to or less than n, the CPU determines "Yes" in step 915 and proceeds to step 920. In step 920, the CPU selects the high-speed targets within the specific range Rs.

[0132] On the other hand, when the number of high-speed targets within the specific range Rs exceeds n, the CPU determines "No" in step 915, proceeds to step 925, and performs a reliability process. After proceeding to step 925, the CPU starts processing from step 1000 (refer to Figure 10 ) and proceeds to step 1005. In step 1005, the CPU calculates the reliability of each high-speed target within the specific range Rs. Next, the CPU proceeds to step 1010 and extracts the targets (high-reliability targets) that satisfy the reliability conditions that hold when the reliability is equal to or greater than the reliability threshold. After that, the CPU temporarily ends the reliability process in step 1095 and proceeds to step 930 (refer to Figure 9A ).

[0133] In step 930, the CPU determines whether the number of high-reliability targets extracted in step 1010 (refer to Figure 10 ) is equal to n. When the number of high-reliability targets is equal to n, the CPU determines "Yes" in step 930 and selects n high-reliability targets in step 940.

[0134] On the other hand, when the number of high-reliability targets is not equal to n, the CPU determines "No" in step 930, proceeds to step 935, and performs a priority process. After proceeding to step 935, the CPU starts processing from step 1100 (refer to Figure 11 ) and proceeds to step 1105. In step 1105, it is determined whether the number of high-reliability targets extracted in the reliability process is less than n. When the number of high-reliability targets is less than n, the CPU determines "Yes" in step 1105 and proceeds to step 1110.

[0135] In step 1110, the CPU determines whether the highly reliable objects are high-speed objects. When the vehicle is going straight, reliability processing is performed on the high-speed objects. Therefore, the CPU determines "yes" in step 1110, proceeds to step 1115, and selects these highly reliable objects. Next, the CPU proceeds to step 1120 and calculates the simple TTC for each of the objects that do not meet the reliability conditions (low-reliability objects). Then, the CPU proceeds to step 1125 and selects the low-reliability objects in ascending order of the simple TTC until the number of selected objects (the sum of the number of highly reliable objects and the number of low-reliability objects) is n. After that, the CPU proceeds to step 1195 and temporarily ends the priority processing.

[0136] In contrast, when the number of highly reliable objects extracted in the reliability processing exceeds n, the CPU determines "no" in step 1105 and proceeds to step 1150. In step 1150, the CPU determines whether the highly reliable objects are high-speed objects. When the vehicle is going straight, the CPU determines "yes" in step 1150 and proceeds to step 1155.

[0137] In step 1155, the CPU calculates the simple TTC for each of the highly reliable objects that exceed n. Then, the CPU proceeds to step 1125 and selects the highly reliable objects in ascending order of the simple TTC until the number of selected objects is n. After that, the CPU proceeds to step 1195 and temporarily ends the priority processing.

[0138] On the other hand, when the vehicle is turning (|r| ≤ r1th), the CPU determines "yes" in step 905 (refer to Figure 9A ) and proceeds to Figure 9B step 945. In step 945, the CPU determines whether the turning angle θ is less than or equal to the angle threshold θth. When θ ≤ θth holds (the turning amplitude is small), the CPU determines "yes" in step 945 and proceeds to step 950.

[0139] In step 950, the CPU determines whether the number of low-speed objects within the specific turning direction range Rt is less than or equal to n. Here, the "low-speed objects within the specific turning direction range Rt" refers to, in other words, the objects that satisfy both the condition "when the object is a low-speed object" and the condition "when the low-speed object exists within the specific turning direction range Rt". When the number of low-speed objects within the specific turning direction range Rt is less than or equal to n, the CPU determines "yes" in step 950 and proceeds to step 960.

[0140] In step 960, the CPU selects low-speed targets within a specific range Rt of the turning direction and proceeds to step 965. In step 965, the CPU determines whether the condition that the number of low-speed targets selected in step 960 is less than n and there are high-speed targets within a specific range Rs holds. When this condition does not hold (i.e., the number of low-speed targets is equal to n or there are no high-speed targets within the specific range Rs), the CPU determines "No" in step 965 and proceeds to step 994 (refer to Figure 9A )(described later). On the other hand, when this condition holds, the CPU determines "Yes" in step 965 and proceeds to step 970 to perform reliability processing.

[0141] After proceeding to step 970, the CPU starts processing from step 1000 (refer to Figure 10 ) and proceeds to step 1005 to calculate the reliability of each high-speed target within the specific range Rs. Next, the CPU proceeds to step 1010 to extract high-reliability targets that meet the reliability condition. After that, the CPU temporarily ends the reliability processing in step 1095 and proceeds to step 975 (refer to Figure 9B ).

[0142] In step 975, the CPU determines whether the number of high-reliability targets extracted in step 1010 (refer to Figure 10 ) is equal to nr (the number obtained by subtracting the number of low-speed targets selected in step 960 from n). When the number of high-reliability targets is equal to nr, the CPU determines "Yes" in step 975 and selects nr high-reliability targets in step 980.

[0143] On the other hand, when the number of high-reliability targets is not equal to nr, the CPU determines "No" in step 975 and proceeds to step 985 to determine whether the number of high-reliability targets is less than nr. When the number of high-reliability targets exceeds nr, the CPU determines "No" in step 985 and proceeds to step 1155 (refer to Figure 11 ).

[0144] In step 1155, the CPU calculates the simple TTC for each of the high-reliability targets exceeding nr. Then, the CPU proceeds to step 1125 and selects high-reliability targets in ascending order of the simple TTC until the number of selected targets (the sum of the number of low-speed targets and the number of high-reliability targets (high-speed targets)) is n. After that, the CPU proceeds to step 1195 and temporarily ends the priority processing.

[0145] On the other hand, when the number of highly reliable landmarks is less than nr, the CPU determines "Yes" in step 985 and selects all the highly reliable landmarks extracted through reliability processing in step 990. Next, the CPU proceeds to step 992 to determine whether there are low-reliability landmarks that do not meet the reliability conditions among the high-speed landmarks within the specific range Rs. When there are no low-reliability landmarks, the CPU determines "No" in step 992.

[0146] On the other hand, when there are low-reliability landmarks, the CPU determines "Yes" in step 992 and proceeds to step 1120 (refer to Figure 11 ) to calculate the simple TTC for each of the low-reliability landmarks. Then, the CPU proceeds to step 1125 and selects the low-reliability landmarks in ascending order of the simple TTC until the number of selected landmarks (the sum of the number of low-speed landmarks, the number of highly reliable landmarks (high-speed landmarks), and the number of low-reliability landmarks (high-speed landmarks)) is n. After that, the CPU proceeds to step 1195 and temporarily ends the priority processing.

[0147] In contrast, when the number of low-speed landmarks within the specific turning direction range Rt exceeds n, the CPU determines "No" in step 950 and starts the reliability processing from step 1000 (refer to Figure 9A ) via step 925 (refer to Figure 10 ) and proceeds to step 1005. The CPU calculates the reliability of each low-speed landmark within the specific turning direction range Rt in step 1005, proceeds to step 1010 to extract the highly reliable landmarks that meet the reliability conditions, and proceeds to step 930 (refer to Figure 9A ) via step 1095.

[0148] In step 930, the CPU determines whether the number of highly reliable landmarks extracted in step 1010 (refer to Figure 10 ) is equal to n. When it is equal to n, the CPU determines "Yes" in step 930 and selects n highly reliable landmarks in step 940.

[0149] On the other hand, when the number of highly reliable landmarks is not n, the CPU determines "No" in step 930 and starts the priority processing from step 1100 (refer to Figure 11 ) via step 935 and proceeds to step 1105. The CPU determines whether the number of highly reliable landmarks extracted in the reliability processing is less than n in step 1105. When it is less than n, the CPU determines "Yes" in step 1105 and determines whether the highly reliable landmarks are high-speed landmarks in step 1110.

[0150] When the vehicle is turning, reliability processing is performed on low-speed targets. Therefore, the CPU determines "No" in step 1110, proceeds to step 1130, and selects these highly reliable targets. Next, the CPU proceeds to step 1135, and based on the front-side target information, obtains the distance from the vehicle for each of the low-reliability targets, and selects these low-reliability targets in order of distance until the number of selected targets (the sum of the number of highly reliable targets and the number of low-reliability targets) is n.

[0151] After that, the CPU proceeds to step 1140 and determines whether the number of selected targets is equal to n. If it is equal to n, the CPU determines "Yes" in step 1140, proceeds to step 1195, and temporarily ends the priority processing.

[0152] On the other hand, when the number of selected targets is not n due to the existence of multiple targets with the same distance, the CPU determines "No" in step 1140, proceeds to step 1145, and sequentially uses the micro-Doppler condition, the speed condition, and the size condition until the number of selected targets is n. After that, the CPU proceeds to step 1195 and temporarily ends the priority processing.

[0153] In contrast, when the number of highly reliable targets extracted in the reliability processing exceeds n, the CPU determines "No" in step 1105 and determines whether the highly reliable targets are high-speed targets in step 1150. When the vehicle is turning, the CPU determines "No" in step 1150 and proceeds to step 1135.

[0154] In step 1135, the CPU obtains the distance from the vehicle for each of the highly reliable targets that exceed n, and selects the highly reliable targets in order of distance until the number of selected targets is n. Next, the CPU proceeds to step 1140 and determines whether the number of selected targets is equal to n. If it is equal to n, it determines "Yes" in step 1140, proceeds to step 1195, and temporarily ends the priority processing. On the other hand, when the number of selected targets is not n, the CPU determines "No" in step 1140 and proceeds to step 1145. The processing of step 1145 is as described above.

[0155] On the contrary, when θ > θth holds (the turning amplitude is large), the CPU determines "No" in step 945 and proceeds to step 955. In step 955, the CPU determines whether the number of low-speed targets within the two-direction specific range Rt is n or less. Here, the "low-speed targets within the two-direction specific range Rt" refers, in other words, to targets that satisfy both the "conditions that hold when the target is a low-speed target" and the "conditions that hold when the low-speed target exists within the two-direction specific range Rt". When the number of low-speed targets within the two-direction specific range Rt is n or less, the CPU determines "Yes" in step 955 and proceeds to step 960. The processing after step 960 is as described above.

[0156] On the other hand, when the number of low-speed targets within the two-direction specific range Rt exceeds n, the CPU determines "No" in step 955 and proceeds to step 925 (see Figure 9A ). The processing after step 925 is as described above.

[0157] After the CPU ends in step 920, 940, or 980, and determines "No" in step 965 or 992 respectively, or after the priority processing in step 1195 ends, the CPU proceeds to step 994 (see Figure 9A ), and excludes non-selected targets. After that, the CPU proceeds to step 995 and temporarily ends this routine (target selection processing), and proceeds to step 820 (see Figure 8 ).

[0158] In step 820, the CPU performs a fusion process of fusing the first front target information, the second front target information, and the front side target information with reduced information amount through the target selection process. Next, the CPU proceeds to step 825, and determines for each of the fused targets whether the collision conditions (alarm conditions and / or automatic braking conditions) hold. When the collision conditions do not hold, the CPU determines "No" in step 825 (that is, determines that there are no targets with a possibility of collision), and proceeds to step 895 and temporarily ends this routine.

[0159] On the other hand, when the collision conditions hold, the CPU determines "Yes" in step 825 and proceeds to step 830. In step 830, when the alarm condition in the collision conditions holds, alarm control is executed as collision avoidance assistance control, and when the automatic braking condition in the collision conditions holds, automatic braking control is executed as collision avoidance assistance control. After that, the CPU proceeds to step 895 and temporarily ends this routine.

[0160] As described above, the collision avoidance assistance device according to the embodiment and the modification has been described, but the present invention is not limited to the above embodiment and modification, and various changes can be made as long as the object of the present invention is not deviated from.

[0161] For example, the ECU 10 is configured to determine that the vehicle is going straight when the vehicle is not turning. In this case, "the case of not turning" is an example of "the first case".

[0162] In addition, instead of or in addition to the front side radar sensors 13L and 13R, the front side target information may be obtained by a front side camera.

[0163] Furthermore, as the collision avoidance assist control, in addition to the alarm control and the automatic braking control, the automatic steering control may be executed. The automatic steering control is a well-known control that automatically changes the steering angle of the steered wheels of the vehicle when the collision condition is satisfied. It should be noted that the alarm control may also be configured to be executed when the alarm condition is satisfied when the vehicle is turning.

[0164] Furthermore, it is not necessarily required to perform the reliability process and / or the priority process. Specifically, it may be configured that when the vehicle is going straight, the targets that satisfy the conditions established when the target is a high-speed target and exists in the specific range Rs are selected (that is, the non-selected targets are excluded). And it may be configured that when the vehicle is turning, the targets that satisfy the conditions established when the target is a low-speed target and exists in the specific range Rt are selected (that is, the non-selected targets are excluded). With this configuration, the amount of the front side target information can also be appropriately reduced.

[0165] Furthermore, in addition to the above cases, it is not necessarily required to set the specific range Rs and the specific range Rt. Specifically, it may be configured that when the vehicle is going straight, the targets that satisfy the conditions established when the target is a high-speed target are selected, and it may also be configured that when the vehicle is turning, the targets that satisfy the conditions established when the target is a low-speed target are selected. With this configuration, the amount of the front side target information can also be appropriately reduced.

[0166] Furthermore, in the above-described embodiment, it is configured that: when the vehicle is turning, low-speed targets existing in a specific range Rt with an upper limit number n are selected, and high-speed targets existing in a specific range Rs are additionally selected with nr as the upper limit only when the number of the low-speed targets is less than n (hereinafter referred to as "Configuration 1"). However, it may also be configured as follows. That is, it may be configured that: when the vehicle is turning, low-speed targets existing in a specific range Rt with an upper limit number n1 (for example, 3) are selected, and high-speed targets existing in a specific range Rs with an upper limit number n2 (n2 = n - n1; for example, 1) are selected (hereinafter referred to as "Configuration 2"). According to Configuration 2, even if there are n or more low-speed targets within the specific range Rt, n2 high-speed targets existing in the specific range Rs can be reliably selected. Therefore, collision avoidance assist control can be appropriately performed even for such high-speed targets. Furthermore, it may also be configured that the driver can switch between Configuration 1 and Configuration 2.

Claims

1. A collision avoidance assistance device, comprising: A forward target information acquisition device that detects a target existing in a specified forward range including at least the front of the own vehicle, and acquires information related to the detected target as forward target information; A front-side target information acquisition device that detects a target existing in a specified front-side range including at least the obliquely front and side of the own vehicle, and acquires information related to the detected target as front-side target information; A vehicle information acquisition device that acquires vehicle information including at least one of the speed of the own vehicle and the yaw angular velocity of the own vehicle or the steering input value of the own vehicle, where the steering input value is an input value based on a steering operation; And A control unit that, when a collision condition that is satisfied when it is determined based on the forward target information and the front-side target information that the own vehicle may collide with the target is established, performs at least one of alarm control for issuing an alarm to the driver of the own vehicle and automatic braking control for automatically applying a braking force to the own vehicle as collision avoidance assistance control, wherein the control unit is configured to: Select a target that satisfies a specified selection condition from the targets included in the front-side target information, and determine for each of the selected targets whether the collision condition is established, The control unit is configured to: When making the determination, determine whether the own vehicle is turning based on the vehicle information, and change the selection condition according to the case where the own vehicle is not turning and the case where the own vehicle is turning, When it is determined that the own vehicle is not turning, further determine whether the own vehicle is going straight, In the first case where the own vehicle is going straight, as the selection condition, it includes a condition that is established when the target is a first target having a speed within a specified first speed range, In the second case where the own vehicle is turning, as the selection condition, it includes a condition that is established when the target is a second target having a speed within a specified second speed range, and the specified second speed range has a lower limit value smaller than the lower limit value of the first speed range and an upper limit value smaller than the upper limit value of the first speed range.

2. The collision avoidance assistance device according to claim 1, wherein, The control unit is configured to: In the first case, as the selection condition, it includes a condition that is established when the first target exists in a specified first specific range within the front-side range, In the second case, as the selection condition, it includes a condition that is established when the second target exists in a specified second specific range narrower than the first specific range within the front-side range.

3. The collision avoidance assistance device according to claim 2, wherein, The vehicle information acquisition device acquires the steering input value, The control unit is configured to: In the second case, Calculate the turning angle by which the own vehicle has turned from the start of turning to the current time point based on the steering input value, When the turning angle exceeds a specified angle threshold, a range including a left range and a right range is set as the second specific range, where the left range includes at least the left front and the left side of the host vehicle, and the right range includes at least the right front and the right side of the host vehicle. When the turning angle is below the angle threshold, the range on the turning direction side of the host vehicle in the left range and the right range is set as the second specific range.

4. The collision avoidance assistance device according to any one of claims 1 to 3, wherein the control unit is configured to: When the number of targets satisfying the selection condition in the first case and the second case respectively exceeds a specified upper limit number, determine whether each target is being continuously detected based on the front side target information, and calculate the reliability of each target based on the determination result. As the selection condition, a reliability condition that is established when the target is a high-reliability target having a reliability above a specified reliability threshold is added.

5. The collision avoidance assistance device according to claim 4, wherein the control unit is configured to: Select targets below the upper limit number from the targets included in the front side target information. the control unit is configured to: In the first case, When the number of high-reliability targets exceeds the upper limit number, for each of the high-reliability targets, calculate a simple collision prediction time specified by the distance from the high-reliability target to the host vehicle and the speed of the high-reliability target itself, and select the high-reliability targets of the upper limit number according to the first priority order in which the smaller the simple collision prediction time, the higher the priority. In the second case, When the number of high-reliability targets exceeds the upper limit number, select the high-reliability targets of the upper limit number according to the second priority order in which the smaller the distance, the higher the priority.

6. The collision avoidance assistance device according to claim 4, wherein the control unit is configured to: Select targets below the upper limit number from the targets included in the front side target information. the control unit is configured to: In the first case, When the number of high-reliability targets is less than the upper limit number, for each of the low-reliability targets that do not satisfy the reliability condition, calculate a simple collision prediction time specified by the distance from the low-reliability target to the host vehicle and the speed of the low-reliability target itself, and select the low-reliability targets in such a way that the sum of the number of high-reliability targets and the number of low-reliability targets is consistent with the upper limit number according to the first priority order in which the smaller the simple collision prediction time, the higher the priority. In the second case, When the number of high-reliability targets is less than the upper limit number, select the low-reliability targets in such a way that the sum of the number of high-reliability targets and the number of low-reliability targets is consistent with the upper limit number according to the second priority order in which the smaller the distance, the higher the priority.

7. The collision avoidance assistance device according to claim 1, wherein the control unit is configured to: In the second case, in addition to the condition that holds when the object is the second object, as the selection condition, it further includes the condition of selecting a specified number of the first objects within a specified first specific range that is less than the specified upper limit number and exists in the front side range.

Images