JP2018097765A - Object detection device and object detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object detection device and an object detection method which can continue to accurately detect the position of an object determined by a fusion target even after the fusion target is no longer generated due to image loss.SOLUTION: When a radar target indicating an object detected by a radar 22 and an image target indicating an object detected by an imaging device 21 are formed by the same object, the radar target and the image target are integrated to form a fusion target. A position, on a vehicle in a width direction, of the image target used for forming the fusion target is calculated as a lateral position, and a moving speed in the width direction of the vehicle is calculated as a lateral speed. When the radar target is acquired while the fusion target is no longer generated due to no acquisition of the image target, a temporal fusion target is formed by using the lateral position and the lateral speed of the image target used for forming the fusion target, and the radar target acquired by a radar target acquisition unit.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an object detection device and an object detection method for detecting an object in front of a host vehicle as a target.

  A technique for increasing the detection accuracy of an object existing around the host vehicle by using a radar and a camera is known. Specifically, a radar target indicating an object detected by the radar is acquired, and an image target indicating the object detected by the camera is acquired. Then, on the condition that the radar target and the image target are determined to be generated from the same object, the radar target and the image target are integrated to form a new target (hereinafter, fusion target). The position of the object is detected by this fusion target.

  However, depending on the traveling environment of the host vehicle, the detection accuracy of the object by the camera may be reduced, and the image may be in a lost state where the object cannot be temporarily detected as an image target. When the image is lost, there is a possibility that a fusion target is not generated for an object in front of the subject vehicle, which is a detection target, and that the vehicle cannot be controlled with respect to the object.

  Therefore, in Patent Document 1, when the image is lost, the size information of the fusion target acquired in the past is multiplied by a coefficient smaller than 1 to obtain the size information of the fusion target acquired in the past. A temporary fusion target having a small size information is generated. And the detection result of the position of an object is updated using this temporary fusion target.

JP 2007-226680 A

  However, Patent Document 1 does not consider the movement of the object in the width direction of the host vehicle. Therefore, for example, if the object is a preceding vehicle and the image is lost due to the direction change of the preceding vehicle, there is a possibility that a deviation may occur between the position of the temporary fusion target and the position of the actual object. there were.

  The present invention has been made in view of the above, and even when the fusion target is no longer generated due to the image lost state, the detection of the position of the object specified by the fusion target is accurately continued. An object of the present invention is to provide an object detection device and an object detection method.

  The present invention includes a radar target acquisition unit (12) that acquires a radar target indicating an object in front of the host vehicle detected by a radar (22) attached in front of the host vehicle, and a front side of the host vehicle. An image target acquisition unit (11) for acquiring an image target indicating the object detected by the imaging device (21) attached to the image sensor, and the radar target and the image target are generated from the same object. A fusion target generation unit (13) for generating a fusion target by integrating the radar target and the image target, and the image target used for generating the fusion target. A horizontal position acquisition unit (14a) that acquires a position of the target in the width direction of the host vehicle as a horizontal position, and a moving speed of the image target used for generating the fusion target in the width direction of the host vehicle. Calculate as lateral speed And a lateral velocity calculation unit (14b) that is used to generate the fusion target when the radar target is acquired and the image target is not acquired and the fusion target is not generated. A temporary fusion target generator (15) that generates a temporary fusion target using the lateral position and velocity of the image target and the radar target acquired by the radar target acquisition unit; A position detection unit (19) for detecting the position of the object using a fusion target or the provisional fusion target.

  According to the present invention, the radar target is acquired, but when the image target is not acquired, the fusion target is not generated. A temporary fusion target is generated using the lateral position and velocity of the image target and the radar target acquired by the radar target acquisition unit, and the position of the object is detected using the temporary fusion target. . In this case, even if the fusion target is no longer generated due to the image lost, it is possible to continue the detection of the position of the object with high accuracy while considering the movement of the object in the lateral direction.

The block diagram of a vehicle control system. The block diagram of ECU. Explanatory drawing regarding the production | generation of a fusion target. The flowchart of the process which a vehicle control system performs. The figure which shows the execution example of the process which a vehicle control system performs.

  Hereinafter, each embodiment will be described with reference to the drawings. In the following embodiments, parts that are the same or equivalent to each other are denoted by the same reference numerals in the drawings, and the description of the same reference numerals is used.

  The vehicle control system 100 according to the present embodiment is a PCS that is mounted on the host vehicle, detects an object existing around the host vehicle as a target, and performs various controls to avoid or reduce collision between the host vehicle and the object. Functions as a (Pre-crash safety system).

  In FIG. 1, the vehicle control system 100 includes an ECU 10, various sensors 20, and a controlled object 30. In the present embodiment, the various sensors 20 include a camera 21, a radar 22, and a vehicle speed sensor 23. The controlled object 30 includes a known safety device such as a speaker, a seat belt, and a brake.

  The camera 21 includes an image sensor such as a CCD or a CMOS, is installed in front of the host vehicle, such as near the upper end of the windshield, and images the front of the host vehicle at a predetermined angle of view. The captured image acquired by the camera 21 is output to the ECU 10 as an image signal.

  The radar 22 is a well-known radar such as a millimeter wave radar, a laser radar, or an ultrasonic radar, and is attached to the front bumper or front grill of the host vehicle so that the optical axis thereof faces the front of the vehicle. The radar 22 transmits radar waves forward from the host vehicle while scanning in a horizontal plane. Then, by receiving the radar wave reflected by the object, the distance, azimuth, relative velocity, etc. with respect to the object reflecting the radar wave are acquired as a radar signal. Then, the acquired radar signal is output to the ECU 10.

  The vehicle speed sensor 23 is provided on a rotary shaft that transmits power to the wheels of the host vehicle, and determines the host vehicle speed that is the speed of the host vehicle based on the rotation speed of the rotary shaft.

  The ECU 10 is an electronic control unit that controls the entire vehicle control system 100. The ECU 10 is mainly composed of a CPU, and includes a ROM, a RAM, an input / output interface, and the like. Various functions to be described below of the ECU 10 are realized by the CPU executing a program stored in the ROM. The history of various data acquired by the ECU 10 is stored in the RAM.

  As shown in FIG. 2, the ECU 10 includes an image target acquisition unit 11, a radar target acquisition unit 12, a fusion target generation unit 13, a lateral position acquisition unit 14 a, a lateral velocity calculation unit 14 b, and a temporary fusion target generation unit 15. , Reliability determination unit 16, crossing determination unit 17, detection state determination unit 18, position detection unit 19, and collision determination unit 40.

  The image target acquisition unit 11 acquires, as an image target GT, an object detected by performing image analysis on the captured image represented by the image signal. For example, the image target GT is acquired by matching processing using a pre-registered target model. The target model is prepared for each type of the image target GT, and the type of the image target GT is specified by this. Examples of the image target GT include obstacles such as a four-wheeled vehicle, a two-wheeled vehicle, a pedestrian, and a guardrail.

  In addition, the image target acquisition unit 11 analyzes the captured image, so that the horizontal position of the image target GT in the width direction of the host vehicle, the size of the image target GT (horizontal width), and the image target GT. Get the direction to the vehicle. Furthermore, the image target acquisition unit 11 applies the position information of the acquired image target GT to the coordinates (XY plane) shown in FIG. 3 to thereby position the image target GT with respect to the host vehicle (detection point Pi). Is identified. The XY plane of FIG. 3 is set with the width direction (lateral direction) of the host vehicle as the X axis and the vehicle length direction (forward direction) of the host vehicle as the Y axis. The tip position (position where the radar 22 is provided) is set as the reference point Po.

  As shown in FIG. 3, the image target acquisition unit 11 sets a detection error region Ri of the image target GT in order to cope with an error included in the detection position of the image target GT. Specifically, an area having a predetermined error width in the vehicle length direction (Y direction) and the lateral direction (X direction) around the detection point Pi (Xi, Yi) of the image target GT is defined as an image object. It is set as the detection error area Ri of the mark GT.

  The radar target acquisition unit 12 acquires an object detected based on the radar signal as a radar target LT. Further, the radar target acquisition unit 12 acquires information on the distance of the radar target LT by analyzing the radar signal. Information on the distance includes a relative distance, a relative speed, a relative acceleration, and the like of the radar target LT. Further, the radar target acquisition unit 12 applies the position information of the acquired radar target LT to the coordinates (XY plane) shown in FIG. 3 to thereby detect the position (detection point Pr) of the radar target LT with respect to the host vehicle. Is identified.

  Further, as shown in FIG. 3, the radar target acquisition unit 12 sets a detection error region Rr of the radar target LT in order to cope with an error included in the detection position of the radar target LT. Specifically, an area having a predetermined error width in the vehicle length direction (Y direction) and the lateral direction (X direction) around the detection point Pr (Xr, Yr) of the radar target LT is set as a radar object. It is set as the detection error region Rr of the mark LT.

  The fusion target generation unit 13 determines whether the image target GT acquired by the image target acquisition unit 11 and the radar target LT acquired by the radar target acquisition unit 12 are generated from the same object. That is, it is determined whether or not the fusion condition is satisfied. In the present embodiment, it is determined that the fusion condition is satisfied when the detection error region Ri and the detection error region Rr in FIG. On the other hand, if there is no overlapping portion between the detection error region Ri and the detection error region Rr, it is determined that the fusion condition is not satisfied.

  When the fusion condition is satisfied, the fusion target generation unit 13 integrates the image target GT and the radar target LT to newly generate a fusion target FT1, and the fusion target FT1 Detect position. At this time, the fusion target FT1 may be generated using information with higher accuracy among the position information included in the image target GT and the radar target LT. That is, the position of the object in the vehicle length direction of the host vehicle is specified using the distance to the radar target LT and the relative speed with respect to the radar target LT. The position of the object in the lateral direction of the host vehicle is specified using the size (lateral width), lateral position, and orientation of the image target GT. As described above, in the example of FIG. 3, the coordinates (Xi, Yr) of the fusion target FT1 are specified by the horizontal position (Xi) of the image target GT and the position (Yr) of the radar target LT in the vehicle length direction. Will be.

  As described above, the fusion target FT1 is used by generating the fusion target FT1 using the information with higher accuracy among the information regarding the radar target LT and the information regarding the image target GT. The detection accuracy of the position of the object can be increased.

  By the way, when the object is a preceding vehicle and the preceding vehicle changes its direction (lane change, right / left turn, etc.), the shape of the rear part of the vehicle captured by the camera 21 changes gradually. Will not be detected. In the following description, from the state in which the fusion target FT1 is detected in this way (fusion state FSN), the image target GT is not detected and the object is detected only by the radar target LT. Is referred to as an “image lost” state.

  However, it is highly possible that the object detected as the fusion target FT1 exists in front of the host vehicle even after the image is lost. Thus, for an object for which the fusion target FT1 is no longer generated due to the image being lost, it is conceivable to update the position information of the object.

  However, in the image lost state, the position of the object in the horizontal direction cannot be specified using the image target GT, and thus the movement of the object in the horizontal direction accompanying the change of direction cannot be detected. Therefore, it is conceivable to specify the position of the object in the lateral direction from the radar target LT. However, the reflection position of the radar 22 is likely to change within the range of the horizontal width of the object, and it is difficult to accurately acquire the horizontal position of the object using the radar target LT. In particular, when the object is changing direction, the reflection position of the radar wave with respect to the object is likely to change.

  Therefore, in the present embodiment, the horizontal position acquisition unit 14a acquires the coordinates (Xi) of the horizontal position of the image target GT used for generating the fusion target FT1. Further, the lateral speed calculation unit 14b calculates the moving speed of the image target in the width direction of the host vehicle as the lateral speed using the history of the lateral position (Xi) of the image target GT acquired by the lateral position acquisition unit 14a. To do. For example, the lateral velocity is calculated from the amount of change per unit time of the lateral position.

  Then, the temporary fusion target generation unit 15 immediately before the image is lost, the lateral position acquired by the lateral position acquisition unit 14a, the lateral speed of the image target GT obtained by the lateral speed calculation unit 14b, and the image The temporary fusion target FT2 is generated using the radar target LT acquired by the radar target acquisition unit 12 after the lost. Note that “immediately before the image is lost” is, for example, a cycle immediately before the cycle in which it is determined that the image has been lost. In addition to this, a period up to a predetermined time (for example, several seconds) before the cycle in which it is determined that the image is lost may be included. As described above, by using the information immediately before the image lost, the detection accuracy of the position of the object after the image lost can be improved. Then, after the image is lost, the position of the object is detected using the temporary fusion target FT2.

  Note that the higher the reliability of the detection result of the object by the fusion target FT1, the higher the possibility that the object exists in front of the host vehicle even after the image is lost. Therefore, the ECU 10 determines the reliability of the detection result of the object by the reliability determination unit 16.

  The reliability determination unit 16 of the present embodiment determines that the reliability of the detection result of the object is high when the number of times the fusion target FT1 is detected before the image is lost is greater than or equal to a predetermined number. The provisional fusion target generation unit 15 generates the provisional fusion target FT2 on the condition that the reliability determination unit 16 determines that the reliability of the object detection result is high.

  Furthermore, when the object is moving in the width direction of the host vehicle, if the moving direction is a direction away from the host vehicle, there is no possibility of the object colliding with the host vehicle. On the other hand, when the object is moving in the width direction of the host vehicle, the object may collide with the host vehicle if the moving direction is a direction approaching the host vehicle. However, even if the object moves in a direction approaching the host vehicle, there is no possibility of the object colliding with the host vehicle when the object crosses (crosses) the front of the host vehicle.

  Therefore, the crossing determination unit 17 is used to determine whether or not the object crosses the front of the host vehicle with the image lost. For example, using the distance and relative speed in the vehicle length direction with respect to the object specified by the radar target LT, and the lateral speed in the vehicle width direction of the object calculated from the image target GT in the fusion state FSN, Determines whether or not the vehicle crosses the front of the host vehicle. Then, the provisional fusion target generation unit 15 generates the provisional fusion target FT2 on the condition that the crossing determination unit 17 determines that the object does not cross the front of the host vehicle.

  Furthermore, if the detection state of the image target GT is unstable, the accuracy of the lateral position and lateral speed of the object acquired from the image target GT may be reduced. For example, when the inter-vehicle distance between the host vehicle and the object is short (in an approaching state), there is a possibility that the lower end of the object will be off from the shooting range of the camera 21. When the lower end is deviated, the detection state of the image target GT becomes unstable, and the accuracy of the lateral position and the lateral speed of the object acquired from the image target GT is lowered. Further, in a situation where it is difficult to photograph the object with the camera 21 such as when a part of the object is hidden behind an obstacle (another object), detection and non-detection of the image target GT are repeated. For example, the detection state of the image target GT may become unstable. Even in this case, the accuracy of the lateral position and the lateral speed acquired from the image target GT is lowered.

  Therefore, the detection state determination unit 18 determines whether or not the detection state of the image target GT used to generate the fusion target FT1 is stable. The provisional fusion target generation unit 15 then acquires the horizontal position and the horizontal position acquired from the image target GT on the condition that the detection state determination unit 18 determines that the detection state of the image target GT is stable. The temporary fusion target FT2 is generated using the speed information. For example, when the lower end of the object is not deviated from the shooting range, it can be determined that the image target GT is stably detected. Further, even when the image target GT is continuously (continuously) detected from the same object, it can be determined that the image target is stably detected. On the other hand, when the lower end is off or when the image target GT is detected intermittently from the same object, it is determined that the detection state of the image target GT is unstable.

  The position detection unit 19 detects the position of the object using the fusion target FT1 or the temporary fusion target FT2. The collision determination unit 40 uses the position information of the object detected by the position detection unit 19 to determine the possibility of collision between the object and the host vehicle.

  When the collision determination unit 40 determines that there is a possibility of a collision, the ECU 10 compares the predicted collision time with the operation timing of each controlled object 30. Here, the collision prediction time is an evaluation value indicating how many seconds later the vehicle collides with an object when traveling at the current host vehicle speed. For example, as the collision prediction time, there is TTC (Time to Collation) calculated by dividing the relative distance between the host vehicle and the object by the relative speed. Further, the collision prediction time includes ETTC (Enhanced Time to Collation) calculated in consideration of relative acceleration in addition to the relative distance and relative speed.

  If the predicted collision time is less than or equal to the operation timing, the corresponding controlled object 30 is operated. For example, if the predicted collision time is less than or equal to the operation timing of the speaker, an alarm is sent to the driver by the operation of the speaker. When the predicted collision time is less than or equal to the seat belt operation timing, control is performed to wind up the seat belt. If the predicted collision time is less than or equal to the brake operation timing, control is performed to reduce the collision speed by operating the automatic brake. As described above, the collision between the host vehicle and the object is avoided or alleviated.

  Next, an execution example of the above process by the ECU 10 will be described with reference to the flowchart of FIG. The following processing is repeatedly performed by the ECU 10 at a predetermined cycle.

  First, the radar target LT is acquired (S11), and the image target GT is acquired (S12). Next, it is determined whether the radar target LT and the image target GT are in the fusion state FSN (S13). In the case of the fusion state FSN (S13: YES), a fusion target FT1 is generated (S14), and information on the vehicle width direction of the image target GT used to generate the fusion target FT1 (lateral position, lateral speed, etc.) ) Is acquired (S15). Next, the position of the object is detected using the fusion target FT1 (S16), and it is determined whether or not there is a possibility of collision between the host vehicle and the object (S17).

  If it is determined in S17 that there is a possibility of collision (S17: YES), it is determined whether or not the predicted collision time TTC is smaller than the operation timing Th1 of the controlled object 30 (S18). If the collision prediction time TTC is smaller than the operation timing Th1 (S18: YES), the controlled object 30 is operated (S19). When it is determined that there is no possibility of collision (S17: NO), and when it is determined that the collision prediction time TTC is longer than the operation timing Th1 of the controlled object 30 (S18: NO), the controlled object 30 is not operated. .

  On the other hand, when it is determined that it is not the fusion state FSN (S13: NO), it is determined whether or not the image is lost (S20). If the image is lost (S20: YES), it is determined whether the reliability of the fusion target FT1 acquired before the image loss is high (S21). When the reliability of the fusion target FT1 is high (S21: YES), it is determined whether or not an elapsed time Tp after the image is lost is shorter than a predetermined threshold Th2 (S22). When the elapsed time Tp is shorter than the threshold Th2 (S22: YES), it is determined whether or not the object crosses the front of the host vehicle (S23). If it is determined that the object does not cross the front of the host vehicle (S23: NO), it is determined whether the detection state of the image target GT is stable (S24). If the image target GT is detected stably (S24: YES), a temporary fusion target FT2 is generated (S25), and the process proceeds to S16.

  As described above, when the process proceeds from S25 to S16, the position of the object is detected using the temporary fusion target FT2. Thereafter, in S17, the collision determination between the host vehicle and the object is performed. Before the provisional fusion target FT2 is generated, the fusion target FT1 is also generated for the same object. Therefore, in S17, the collision determination may be performed using the position information calculated from the movement trajectories of the fusion target FT1 and the temporary fusion target FT2.

  When it is determined that the image is not lost (S20: NO), when it is determined that the reliability of the fusion target FT1 acquired before the image lost is low (S21: NO), the elapsed time Tp is longer than the threshold Th2. In the case (S22: NO), when it is determined that the object crosses the front of the host vehicle (S23: YES), the detection state of the image target GT acquired before the image is lost is unstable (S24). : NO), the process is terminated. In these cases, the provisional fusion target FT2 is not generated, and the detection of the position of the object ends.

  Next, an execution example of the above process will be described with reference to FIG. FIG. 5 shows a state where the host vehicle M1 and the preceding vehicle M2 are both traveling forward of the vehicle, and the preceding vehicle M2 is making a right turn.

  First, in the period P1, when the preceding vehicle M2 is detected by both the radar 22 and the camera 21, the position of the preceding vehicle M2 is detected by the fusion target FT1. During this period P1, information (lateral position, lateral speed, etc.) regarding the vehicle width direction of the preceding vehicle M2 is acquired using the image target GT. Further, a collision determination with the host vehicle M1 is performed using the fusion target FT1.

  Thereafter, when the preceding vehicle M2 decelerates and makes a right turn, the relative position with respect to the host vehicle M1 changes in time series as illustrated. Then, with the right turn of the preceding vehicle M2, the camera 21 is in an image lost state at a timing A at which the camera 21 cannot catch the rear end portion of the preceding vehicle M2. When the image is lost, the lateral speed of the preceding vehicle M2 is calculated based on the history of the lateral position of the image target GT used to generate the fusion target FT1.

  Thereafter, in the period P2 in the image lost state, the radar target LT detected by the radar 22 in the period P2, the lateral speed of the preceding vehicle M2 obtained from the history of the image target GT in the period P1, and the period The temporary fusion target FT2 is generated using the lateral position of the preceding vehicle M2 obtained from the image target GT acquired immediately before the image lost in P1. Then, the current position of the preceding vehicle M2 is detected by the temporary fusion target FT2, and the collision determination is performed using the temporary fusion target FT2.

  According to the above, the following excellent effects can be achieved.

  The object in front of the host vehicle detected by the fusion target FT1 exists in front of the host vehicle even if the image target GT is not detected and only the radar target LT is detected. Therefore, it is preferable that the detection of the position of the object is continued. However, in the image lost state, the position of the object in the width direction of the host vehicle cannot be specified with high accuracy. Therefore, the position in the width direction of the host vehicle of the image target GT used for generating the fusion target FT1 is acquired as the lateral position. Further, the moving speed of the image target GT used for generating the fusion target FT1 in the width direction of the host vehicle is acquired as the lateral speed. In the image lost state, the radar object acquired by the radar 22 after the lateral position and velocity of the image target GT used for generating the fusion target and the fusion target FT1 are not generated (after the image lost). The temporary fusion target FT2 is generated using the target LT, and the position of the object is detected using the temporary fusion target FT2. In this case, even if the fusion target FT1 is no longer generated, the position of the object can be detected accurately and continuously using the temporary fusion target FT2 in consideration of the lateral movement of the object. It becomes possible.

  -Since the provisional fusion target FT2 is generated on the condition that the reliability of the detection result of the object by the fusion target FT1 is high, in the state where the image is lost, it is limited to the object with high accuracy existing in front of the host vehicle. Thus, the detection of the position of the object can be continued.

  The reliability of the detection result of the object can be determined by utilizing the fact that the probability that the object exists ahead of the host vehicle increases as the number of detections of the fusion target FT1 increases.

  If the movement of the object detected as the fusion target FT1 in the width direction of the host vehicle is a direction approaching the host vehicle and crosses the front of the host vehicle, the temporary fusion target FT2 is not generated. It was to be. As described above, after the image is lost, the provisional fusion target FT2 is generated for an object that is not likely to collide with the host vehicle, thereby suppressing the occurrence of inconvenience such as unnecessary control. Can do.

  Even if the fusion target FT1 has been generated, if the detection state (recognition) of the image target GT is unstable, the reliability of the calculation results such as the lateral position and the lateral speed of the object may be reduced. . Therefore, the provisional fusion target FT2 is generated on the condition that the detection state of the image target GT is stable. As described above, it can be suppressed that the detection accuracy of the position of the object by the temporary fusion target FT2 is lowered due to the unstable detection state of the image target GT.

  The lateral velocity of the object can be obtained using the history of the lateral position of the image target GT.

  The above embodiment may be modified as follows, for example. In the following description, the same components as those described above are denoted by the same reference numerals, and detailed description thereof is omitted.

  -In above-mentioned FIG. 2, you may abbreviate | omit each process of the reliability determination part 16, the crossing determination part 17, and the detection state determination part 18 of ECU10. For example, the ECU 10 may generate the temporary fusion target FT2 regardless of the processes 16 to 18 when the image lost state is entered.

  In FIG. 2, regardless of each of the determination units 16 to 18, if the elapsed time after the image lost exceeds a predetermined time, the accuracy of the presence of the object is lowered, and therefore the provisional fusion target It is preferable not to generate FT2. In this case, it is possible to suppress the occurrence of inconvenience such as unnecessary control by generating the temporary fusion target FT2 in a situation where the accuracy of the existence of the object is low.

  In FIG. 2, the reliability determination unit 16 variably sets the period P2 for generating the temporary fusion target FT2 according to the number of detections of the fusion target FT1 before the image is lost. There may be. That is, as the number of detections of the fusion target FT1 increases, the period for generating the temporary fusion target FT2 is set longer. As described above, after the image is lost, the detection of the object can be continuously performed according to the accuracy with which the object exists.

  -In above-mentioned FIG. 2, the reliability determination part 16 may determine a reliability according to the period when the fusion target FT1 was detected. That is, it is determined that the reliability of the detection result of the object is higher as the detection period of the fusion target FT1 is longer. Also in this case, when the image is lost, the detection of the object can be continuously performed according to the accuracy with which the object exists.

  In FIG. 2 described above, the reliability determination unit 16 may determine that the reliability is low when the elapsed time after image loss exceeds a predetermined time. In this case, it is possible to suppress the occurrence of inconveniences such as unnecessary control by continuously generating the temporary fusion target FT2 in a situation where the probability that an object exists is low.

  In FIG. 2, the reliability determination unit 16 determines the reliability of the detection result of the object based on the degree of coincidence between the radar target LT and the image target GT when the fusion condition is satisfied. May be. For example, in FIG. 3, it is determined that the degree of coincidence between the radar target LT and the image target GT increases as the overlapping portion between the detection error region Ri of the image target GT and the detection error region Rr of the radar target LT increases. . Then, as the degree of coincidence between the radar target LT and the image target GT increases, the period for generating the temporary fusion target FT2 is set longer. In this case, the reliability of the detection result of the object can be determined using the fact that the higher the degree of coincidence between the radar target LT and the image target GT, the higher the probability that the object exists.

  The ECU 10 includes a lateral acceleration calculation unit that calculates acceleration (lateral acceleration) in the vehicle width direction (lateral direction) of the object using the image target GT used to generate the fusion target FT1. The fusion target generation unit 15 may generate the temporary fusion target FT2 in consideration of the lateral acceleration calculated by the lateral acceleration calculation unit. The lateral acceleration can be calculated as a change amount of the lateral velocity calculated by the lateral velocity calculation unit 14b.

  By considering the lateral acceleration of the image target GT, the temporary fusion target FT2 can be generated in consideration of the approach state between the host vehicle and the object, and the lateral position of the object by the temporary fusion target FT2 can be generated. Calculation accuracy can be increased. However, in this case, if the direction (vector) of the lateral velocity and the lateral acceleration do not match, the lateral position of the object may not be obtained correctly. Therefore, the provisional fusion target generation unit 15 may generate the provisional fusion target FT2 using the lateral velocity and the lateral acceleration on the condition that the direction of the lateral velocity and the lateral acceleration coincide with each other. By generating the temporary fusion target FT2 on the condition that the own vehicle and the object are in an approaching state and the lateral velocity and the lateral acceleration of the target coincide with each other, the possibility of collision between the own vehicle and the object is high. In this case, the position of the object can be continuously detected using the temporary fusion target FT2.

  In FIG. 2 described above, the provisional fusion target generation unit 15 does not generate the provisional fusion target FT2 when the crossing determination unit 17 determines that the object crosses the front of the host vehicle. However, if the distance in the traveling direction between the object and the host vehicle becomes shorter at the timing when the object crosses the front of the host vehicle, some warning may be given to the object. Therefore, when the crossing determination unit 17 determines that the object crosses the front of the host vehicle, the temporary fusion target generation unit 15 determines that the distance between the object and the host vehicle when the object crosses the front of the host vehicle is less than a predetermined value. In such a case, the determination result of the crossing determination unit 17 is invalidated. That is, even if it is determined that the object crosses the front of the host vehicle, the generation of the temporary fusion target FT2 may be permitted. In this case, the threshold value of the distance between the object and the host vehicle when the object crosses the front of the host vehicle may be variably set according to the approach state (relative speed) between the host vehicle and the object.

  -In the above, even when the host vehicle avoids a stopped vehicle ahead by steering, there is a possibility that the image is lost. Therefore, the ECU 10 may estimate the curve radius of the host vehicle estimated from the steering angle of the host vehicle or the yaw rate sensor output, and correct the lateral position of the object in the image lost state.

  DESCRIPTION OF SYMBOLS 11 ... Image target acquisition part, 12 ... Radar target acquisition part, 13 ... Fusion target production | generation part, 14a ... Lateral position acquisition part, 16 ... Reliability determination part, 19 ... Position detection part, 21 ... Camera.

Claims (12)

A radar target acquisition unit (12) for acquiring a radar target indicating an object in front of the host vehicle detected by a radar (22) mounted in front of the host vehicle;
An image target acquisition unit (11) for acquiring an image target indicating the object detected by the imaging device (21) attached in front of the host vehicle;
A fusion target generation unit that generates a fusion target by integrating the radar target and the image target when the radar target and the image target are generated from the same object. (13)
A lateral position acquisition unit (14a) that acquires a position in the width direction of the host vehicle of the image target used for generating the fusion target as a lateral position;
A lateral speed calculation unit (14b) that calculates, as a lateral speed, a moving speed of the image target used for generating the fusion target in the width direction of the host vehicle;
When the radar target is acquired and the image target is not acquired, and thus the fusion target is not generated, the lateral position of the image target used for generating the fusion target and A temporary fusion target generating unit (15) that generates a temporary fusion target using the lateral velocity and the radar target acquired by the radar target acquiring unit;
A position detector (19) for detecting the position of the object using the fusion target or the temporary fusion target;
An object detection apparatus comprising: A reliability determination unit (16) for determining the reliability of the detection result of the object by the fusion target;
The object detection device according to claim 1, wherein the temporary fusion target generation unit generates the temporary fusion target on the condition that the reliability is determined to be high.   The object detection apparatus according to claim 2, wherein the reliability determination unit determines that the reliability is higher as the number of detections of the fusion target is larger.   The reliability determination unit is specified by the position of the object specified by the radar target used for generating the fusion target and the image target used for generating the same fusion target. The object detection apparatus according to claim 2, wherein the higher the degree of coincidence with the position of the object, the higher the reliability is determined.   The object detection device according to claim 2, wherein the temporary fusion target generation unit sets a period for generating the temporary fusion target to be longer as the reliability is higher.   The temporary fusion target generation unit does not generate the temporary fusion target when the elapsed time since the fusion target is no longer generated exceeds a predetermined time. The object detection apparatus described. Crossing for determining whether or not the object crosses the front of the host vehicle when the movement of the object detected as the fusion target in the width direction of the host vehicle is a direction approaching the host vehicle. A determination unit (17);
The object detection according to claim 1, wherein the provisional fusion target generation unit does not generate the provisional fusion target when it is determined that the object crosses the front of the host vehicle. apparatus. A detection state determination unit (18) for determining whether or not the detection state of the image target is stable;
The said temporary fusion target production | generation part produces | generates the said temporary fusion target on the conditions that it determined with the detection state of the said image target being stable. Object detection device.   9. The object detection device according to claim 1, wherein the lateral velocity calculation unit acquires a lateral velocity of the object based on a history of acquisition of the lateral position by the lateral position acquisition unit. A lateral acceleration calculation unit (10) for obtaining a lateral acceleration which is an acceleration when the image target used for generating the fusion target moves in the width direction of the host vehicle;
The object detection device according to claim 1, wherein the temporary fusion target generation unit generates the temporary fusion target using the lateral acceleration of the image target.   The temporary fusion target generation unit generates the temporary fusion target on the condition that the directions of the lateral velocity and the lateral acceleration coincide and the object and the host vehicle are in an approaching state. The object detection apparatus according to claim 10. Obtaining a radar target indicative of an object ahead of the host vehicle detected by a radar (22) mounted in front of the host vehicle;
Obtaining an image target indicating the object detected by the photographing device (21) attached in front of the host vehicle;
When the radar target and the image target are generated from the same object, the step of integrating the radar target and the image target to generate a fusion target;
Obtaining a position in the width direction of the host vehicle of the image target used for generating the fusion target as a lateral position;
Acquiring a moving speed of the image target used for generating the fusion target in the width direction of the host vehicle as a lateral speed;
When the radar target is acquired and the image target is not acquired, and thus the fusion target is not generated, the lateral position of the image target used for generating the fusion target and Generating a temporary fusion target using the lateral velocity and the radar target acquired after the fusion target is no longer generated;
Detecting the position of the object using the fusion target or the temporary fusion target;
An object detection method comprising:

Images