JP2010108168A - Collision prediction device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a collision prediction device for accurately determining the probability that a target object may collide with one's own vehicle according to the detecting accuracy of an image sensor when the target object is detected by means of a radar sensor and the image sensor. <P>SOLUTION: A collision prediction ECU 1 includes a collision probability calculating unit 18 and a collision probability correcting and computing unit 20. The collision probability calculating unit 18 calculates the position and probability of collision of the one's own vehicle with an incoming vehicle on the basis of the track and the vehicle speed of the one's own vehicle and the vehicle speed and the track of the incoming vehicle. The collision probability correcting and computing unit 20 detects the relative positional relationship between a lateral position of the incoming vehicle and an edge of the incoming vehicle, and corrects the probability of the collision based on the relative positional relationship between the lateral position of the incoming vehicle and the edge of the incoming vehicle. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a collision prediction apparatus that predicts a collision probability with a target object around a host vehicle such as another vehicle in the host vehicle.

Conventionally, there is a vehicle detection device using a radar sensor and an image sensor as a device for detecting a vehicle around the host vehicle (for example, Patent Document 1). This vehicle detection device recognizes a forward vehicle by clustering processing of each reflection point received by the forward search radar, extracts an edge by image processing on an image obtained by capturing the front of the host vehicle, and moves forward based on the extracted edge. Recognize the vehicle. Sensor fusion is performed on the preceding vehicle recognized by these clustering processing and image processing, and the preceding vehicle is detected by this sensor fusion. Then, the possibility of a collision between the detected preceding vehicle and the host vehicle is determined.
JP 2005-141517 A

  However, in the vehicle detection device disclosed in Patent Document 1, when extracting an edge by image processing, the distance to the extracted edge cannot be acquired, and when a plurality of edges are extracted, the edge and the front In some cases, the relationship with the vehicle was unclear. For this reason, in spite of detecting the edge of the object in front of the vehicle, it becomes unclear whether the detected edge is that of the preceding vehicle, and as a result, the detection accuracy of the preceding vehicle is lowered. And since the detection accuracy of the vehicle ahead becomes low, there is a problem that the determination accuracy of the possibility of collision is low.

  Therefore, an object of the present invention is to provide a collision prediction device that can accurately determine the collision probability of a target object with respect to the host vehicle according to the detection accuracy of the image sensor when the target object is detected by the radar sensor and the image sensor. It is to provide.

  A collision prediction apparatus according to the present invention that solves the above problems includes a radar detection unit that detects a target object using a radar, an image that detects an edge from an image captured by the imaging unit, and detects the target object based on the detected edge. A collision probability calculating unit for calculating a collision probability between the target object and the host vehicle based on a relationship between the detection unit, the target object and the host vehicle; and a target object detected by the radar detecting unit with respect to the vehicle width direction in the host vehicle. A relative positional relationship detecting means for detecting a relative positional relationship between a radar lateral position as a position and an edge width direction position at a target object corresponding edge that is an edge corresponding to the target object detected by the image detecting means; A collision probability correction method for correcting the collision probability between the target object and the host vehicle based on the relative positional relationship between the edge corresponding to the target object and the position in the edge width direction. Characterized in that it comprises a and.

  The collision prediction apparatus according to the present invention calculates the collision probability between the target object and the host vehicle based on the relationship between the target object and the host vehicle. Here, even if the detection accuracy of the target object-corresponding edge is low, it is considered that the detection accuracy of the target object-corresponding edge depends on the relative positional relationship between the radar lateral position and the width direction position of the target object-corresponding edge. In this regard, in the collision prediction apparatus according to the present invention, the relative position of the radar lateral position, which is the position of the target object with respect to the vehicle width direction of the host vehicle, and the edge width direction position of the target object corresponding edge that is the edge corresponding to the target object. Based on the positional relationship, the collision probability between the target object and the host vehicle is corrected. For this reason, when the target object is detected by the radar sensor and the image sensor, the collision probability of the target object with respect to the host vehicle can be accurately determined according to the detection accuracy of the image sensor.

  Here, the collision probability correcting means is configured such that when the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction, compared to the case where the radar lateral position is located at the center of the target object corresponding edge in the edge width direction. It can be set as the aspect correct | amended so that a probability may become relatively low.

  If the radar lateral position is located at the center of the edge corresponding to the target object in the edge width direction, the collision probability is considered to be higher than if the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction. . Therefore, in the collision prediction device according to the present invention, when the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction, compared to the case where the radar lateral position is located at the center of the target object corresponding edge in the edge width direction. The correction is made so that the collision probability is relatively low. For this reason, it is possible to accurately determine the collision probability of the target object with respect to the host vehicle.

  Note that in the present invention, “when the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction, the collision probability is lower than that in the center of the target object corresponding edge in the edge width direction. "Correction to be so that" means that the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction with reference to the case where the radar lateral position is located at the center of the target object corresponding edge. If the radar lateral position is located at the edge width direction edge of the target object corresponding edge, the radar lateral position is the center in the edge width direction of the target object corresponding edge. The case where the collision probability when positioned in a part is corrected to be high is included. Furthermore, “When the radar lateral position is located at the center in the width direction of the edge corresponding to the target object, the radar lateral position is located between the edge in the edge width direction and the center in the edge corresponding to the target object. Includes a case where the collision probability is corrected to be high and the collision probability is corrected to be low when positioned at the end in the width direction.

  In addition, the collision probability correcting means is far from the own vehicle in the edge width direction at the target object corresponding edge when the radar lateral position is located at the end near the own vehicle in the edge width direction at the target object corresponding edge. It can also be set as the aspect correct | amended so that a collision probability may become relatively high compared with the case where it is located in the edge part of the side.

  When the target object may cause an offset collision with respect to the host vehicle, the target object-corresponding edge may have an end portion closer to the host vehicle and an end portion farther from the host vehicle. Also, when an object located on the side of the vehicle is misrecognized as being integrated with the target object, the target object-corresponding edge has an end near the host vehicle and an end far from the host vehicle. There is a case. Here, when the target object may cause an offset collision with respect to the host vehicle, the radar lateral position is located at the end of the target object corresponding edge on the side closer to the host vehicle in the edge width direction. On the other hand, when the object located on the side of the vehicle is erroneously recognized as being integrated with the target object, the position in the edge width direction at the target object corresponding edge at the lateral position of the radar is considered to be various positions. Therefore, in the collision prediction device according to the present invention, when the radar lateral position is located at the end of the target object corresponding edge near the own vehicle in the edge width direction, When located at the far end, the collision probability is corrected to be relatively high. For this reason, when the target object may cause an offset collision with respect to the host vehicle, it is possible to prevent the collision probability from being estimated too low. Therefore, the collision probability of the target object with respect to the host vehicle in the target object can be determined with higher accuracy.

  In the present invention, when the radar lateral position is located at the end of the target object corresponding edge near the host vehicle in the edge width direction, “Correction so that the collision probability is relatively higher than when positioned at the end” means that “the radar lateral position is at the end of the target object corresponding edge in the edge width direction far from the vehicle. When the position of the radar is positioned, the radar lateral position is corrected to be higher when the edge corresponding to the target object is located at the end of the edge width direction closer to the host vehicle. When the radar lateral position is located at the end of the target object corresponding edge that is far from the host vehicle in the edge width direction, based on the case where the edge is located at the end near the own vehicle in the edge width direction Collision probability to the low correction "including the case. Furthermore, “If the radar lateral position is located at the edge of the edge corresponding to the target object, the center of the edge corresponding to the edge width direction is the reference, the collision probability In the edge width direction, the collision probability is corrected to be low.

  Furthermore, the vehicle has a lateral velocity detecting means for detecting whether the target object has a lateral velocity component that is a velocity component along a direction orthogonal to the traveling direction of the host vehicle, and detects the collision probability. When the target object is approaching with a lateral velocity component, the correcting means is subject to the subject vehicle in the edge width direction at the target object corresponding edge as compared with the case where the target object is not approaching with the lateral velocity component. It is possible to adopt a mode in which the collision probability is corrected so that the collision probability is relatively high when it is located at the far end.

  When the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction far from the host vehicle, the target object has a lateral velocity component and approaches the host vehicle, In some cases, a stationary object may be erroneously detected as a target object. Here, if the detected object has a lateral velocity component, the target object is highly likely to collide with the host vehicle. If the detected object does not have a lateral velocity component, the host vehicle collides. It is considered that an object with a low possibility is erroneously detected as a target object. In this regard, in the collision prediction device according to the present invention, when the target object has a lateral velocity component and approaches the host vehicle, the target object has a lateral velocity component and does not approach. Thus, the collision probability when the edge corresponding to the target object is located at the end far from the host vehicle in the edge width direction is corrected so as to be relatively high. For this reason, it is possible to prevent the collision probability of the target object with respect to the own vehicle from being estimated too low when the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction far from the own vehicle. As a result, the collision probability of the target object with respect to the host vehicle can be determined with higher accuracy.

  Note that in the present invention, “when the target object is approaching with a lateral velocity component, the edge width direction at the target object corresponding edge compared to when the target object is not approaching with a lateral velocity component. `` Correcting the collision probability so that the collision probability is relatively high when it is located at the end far from the host vehicle '' means that the target object has a lateral velocity component and is not approaching. As a reference, when the target object is approaching with a lateral velocity component, the collision probability when the edge corresponding to the target object is located at the end of the edge width direction far from the host vehicle is corrected. "When the target object is approaching with a lateral velocity component, the vehicle in the edge width direction at the target object corresponding edge when the target object is not approaching with a lateral velocity component." If it is located at the end far from the Probability of correcting lower, "including the case.

  In addition, a collision prediction apparatus according to the present invention that has solved the above-described problem is a radar detection unit that detects a target object using a radar, a plurality of edges detected from an image captured by the imaging unit, and based on the detected plurality of edges. Collision probability calculation that calculates the collision probability between the target object and the host vehicle based on the relationship between the target object detected by the image detection unit that detects the target object, the target object detected by the radar detection unit and the image detection unit, and the host vehicle. And a collision probability correction unit that corrects the collision probability according to the edge accuracy, and an edge accuracy detection unit that detects an edge accuracy that is an accuracy of a plurality of edges belonging to the same object. It is.

  The collision prediction apparatus according to the present invention corrects the collision probability according to the edge accuracy. For this reason, the collision probability when an edge different from the target object is detected can be lowered. Therefore, when the target object is detected by the radar sensor and the image sensor, the collision probability of the target object with respect to the own vehicle is accurately determined. can do.

  According to the collision prediction apparatus according to the present invention, when the target object is detected by the radar sensor and the image sensor, the collision probability of the target object with respect to the host vehicle can be accurately determined according to the detection accuracy by the image sensor.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted. For the convenience of illustration, the dimensional ratios in the drawings do not necessarily match those described.

  FIG. 1 is a block configuration diagram showing a configuration of a collision prediction apparatus according to the present invention. As shown in FIG. 1, the collision prediction apparatus includes a collision prediction ECU 1. A steering angle sensor 2, a yaw rate sensor 3, a wheel speed sensor 4, a millimeter wave radar sensor 5, and an image sensor 6 are connected to the collision prediction ECU 1. In addition, a collision avoidance device such as an alarm device 31, a collision avoidance support device 32, a seat belt control device 33, a seat position control device 34, a brake control device 35, and an airbag control device 36 is connected to the collision prediction ECU 1. .

  The collision prediction ECU 1 is a computer of an automobile device that is electronically controlled, and includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), an input / output interface, and the like. The collision prediction ECU 1 includes an estimated curve radius calculation unit 11, a host vehicle track calculation unit 12, a host vehicle speed calculation unit 13, and a counterpart vehicle speed calculation unit 14. The collision prediction ECU 1 includes an opponent vehicle movement distance calculation unit 15, an opponent vehicle lateral position calculation unit 16, and an opponent vehicle track calculation unit 17. The collision prediction ECU 1 further includes a collision probability calculation unit 18, an image processing unit 19, and a collision probability correction / calculation unit 20. Further, the collision probability calculation unit 18 is provided with a collision position calculation unit 21 and a collision position distribution calculation unit 22.

  The steering angle sensor 2 is attached to a steering rod in a vehicle, for example, and detects the steering angle of the steering wheel operated by the driver. The steering angle sensor 2 outputs a steering angle signal related to the detected steering angle to the estimated curve radius calculation unit 11 in the collision prediction ECU 1.

  The yaw rate sensor 3 is provided, for example, at the center position of the vehicle body, and detects the yaw rate applied to the vehicle body. The yaw rate sensor 3 outputs a yaw rate signal related to the detected yaw rate to the estimated curve radius calculation unit 11 in the collision prediction ECU 1.

  The wheel speed sensor 4 is attached to a wheel portion of the vehicle, for example, and detects the wheel speed of the vehicle. The wheel speed sensor 4 outputs a wheel speed signal related to the detected wheel speed to the host vehicle speed calculation unit 13 in the collision prediction ECU 1.

  The millimeter wave radar sensor 5 is attached to, for example, the front grille and rear trunk cover portions of the vehicle, emits millimeter waves forward and sideward, and receives the reflected waves. The millimeter wave radar sensor 5 outputs the reflected wave signal relating to the received reflected wave to the opponent vehicle speed calculation unit 14, the opponent vehicle movement distance calculation unit 15, and the opponent vehicle lateral position calculation unit 16 in the collision prediction ECU 1. The millimeter wave radar sensor 5 constitutes radar detection means.

  The image sensor 6 includes an in-vehicle camera that images the front of the vehicle. The image sensor 6 transmits an image ahead of the vehicle imaged by the in-vehicle camera to the image processing unit 19 in the collision prediction ECU 1. When an obstacle such as another vehicle traveling in front of the vehicle is present, the obstacle is reflected in an image captured by the in-vehicle camera. The image sensor 6 constitutes image detection means.

  Based on the steering angle signal output from the steering angle sensor 2 and the yaw rate signal output from the yaw rate sensor 3, the estimated curve radius calculation unit 11 in the collision prediction ECU 1 calculates the estimated curve radius of the host vehicle by calculation. The estimated curve radius calculation unit 11 outputs the calculated estimated curve radius to the host vehicle track calculation unit 12.

  The host vehicle track calculation unit 12 calculates the track of the host vehicle by calculation based on the estimated curve radius signal output from the estimated curve radius calculation unit 11. The host vehicle track calculation unit 12 outputs the host vehicle track that is the calculated track of the host vehicle to the collision probability calculation unit 18.

  The own vehicle speed calculation unit 13 calculates the vehicle speed of the own vehicle by calculation based on the wheel speed signal output from the wheel speed sensor 4. The host vehicle speed calculation unit 13 outputs the calculated host vehicle speed, which is the vehicle speed of the host vehicle, to the collision probability calculation unit 18.

  The counterpart vehicle speed calculation unit 14 calculates the vehicle speed of the counterpart vehicle, which is a moving body, based on the reflected wave signal output from the millimeter wave radar sensor 5 by calculation. The partner vehicle speed calculation unit 14 outputs the partner vehicle speed, which is the calculated vehicle speed of the partner vehicle, to the collision probability calculation unit 18.

  Based on the reflected wave signal output from the millimeter wave radar sensor 5, the opponent vehicle movement distance calculation unit 15 calculates the movement distance of the opponent vehicle after the millimeter wave radar sensor 5 captures the opponent vehicle. The partner vehicle movement distance calculation unit 15 outputs the partner vehicle movement distance, which is the calculated movement distance of the partner vehicle, to the partner vehicle trajectory calculation unit 17.

  The partner vehicle lateral position calculation unit 16 calculates the partner vehicle lateral position, which is the position of the partner vehicle in the vehicle width direction of the host vehicle, based on the reflected wave signal output from the millimeter wave radar sensor 5. The opponent vehicle lateral position calculation unit 16 outputs the calculated opponent vehicle lateral position to the opponent vehicle track calculation unit 17 and the collision probability correction / calculation unit 20.

  The partner vehicle trajectory calculation unit 17 calculates the trajectory of the partner vehicle based on the partner vehicle lateral position output from the partner vehicle lateral position calculator 16 and the partner vehicle travel distance output from the partner vehicle travel distance calculator 15. calculate. The opponent vehicle trajectory calculation unit 17 outputs the opponent vehicle trajectory that is the calculated trajectory of the opponent vehicle to the collision probability calculation unit 18.

  In the collision position calculation unit 21, the collision probability calculation unit 18 includes a host vehicle track output from the host vehicle track calculation unit 12, a host vehicle speed output from the host vehicle speed calculation unit 13, and a partner output from the partner vehicle speed calculation unit 14. Based on the vehicle speed and the opponent vehicle track output from the opponent vehicle track calculation unit 17, the collision position of the opponent vehicle in the host vehicle is calculated by calculation.

  In addition, the collision probability calculation unit 18 calculates the collision position distribution in the collision position distribution calculation unit 22. In addition, the collision position distribution calculation unit 22 divides the length of the host vehicle in the width direction into a plurality of sections in advance at predetermined intervals, and which section includes the collision position calculated by the collision position calculation unit 21. to decide.

  Subsequently, when the section including the collision position is determined, the collision position count value is added to the section including the collision position. Thus, for each section, the number of times calculated by the collision position calculation unit 21 is counted, and the counting result is calculated as a collision position distribution. In addition, when the number of counts reaches a predetermined number, the collision position distribution calculation unit 22 deletes old collision position data and counts new collision position data.

  Furthermore, the collision probability calculation unit 18 calculates the collision probability of the opponent vehicle with respect to the host vehicle based on the calculated collision position distribution. Here, the collision probability calculation unit 18 calculates a collision probability as a probability of collision with each partitioned section. The collision probability calculation unit 18 outputs the calculated collision probability for each section to the collision probability correction / calculation unit 20.

  Further, the collision probability calculation unit 18 determines the own vehicle trajectory output from the own vehicle trajectory calculation unit 12, the opponent vehicle trajectory output from the opponent vehicle track calculation unit 17, and the opponent vehicle speed output from the opponent vehicle speed calculation unit 14. Based on this, the opponent vehicle lateral speed which is the lateral speed of the opponent vehicle with respect to the own vehicle is calculated. The collision probability calculation unit 18 outputs the calculated opponent vehicle lateral speed to the collision probability correction / calculation unit 20. The collision probability calculation unit 18 constitutes a collision probability calculation unit.

  The image processing unit 19 performs edge processing by performing image processing on the image transmitted from the image sensor 6. Further, the image processing unit 19 detects an opponent vehicle edge corresponding to the opponent vehicle based on the detected edge. Further, the image processing unit 19 outputs the detected opponent vehicle edge to the collision probability correction / calculation unit 20.

  The collision probability correction / calculation unit 20 corresponds to the collision probability output from the collision probability calculation unit 18 and the partner vehicle lateral position output from the partner vehicle lateral position calculation unit 16 and the partner vehicle output from the image processing unit 19. The relative positional relationship with the edge is detected. The collision probability correction / calculation unit 20 constitutes a relative positional relationship detection unit.

  The collision probability correction / calculation unit 20 corrects the collision probability based on the detected relative positional relationship between the opponent vehicle lateral position and the opponent vehicle edge. The collision probability correction / calculation unit 20 constitutes a collision probability correction unit. Further, the opponent vehicle lateral velocity output from the collision probability calculation unit 18 and the edge accuracy which is the accuracy that the opponent vehicle edge belongs to the same object are detected, and the collision probability is corrected using the opponent vehicle lateral velocity and edge accuracy. . The collision probability correction / calculation unit 20 constitutes edge accuracy detection means.

  The collision probability correction / calculation unit 20 predicts the collision position of the own vehicle with the opponent vehicle based on the corrected collision probability. The collision probability correction / calculation unit 20 outputs a collision response operation signal corresponding to the calculated collision probability relating to the collision position to an alarm device 31, which is a collision avoidance device, a collision avoidance support device 32, a seat belt control device 33, and a seat position control device 34. , Output to the brake control device 35 and the airbag control device 36, respectively.

  The alarm device 31 includes a speaker that emits a warning sound, a speaker control device that controls the speaker, a monitor that displays an alarm, and a monitor control device. Based on the collision response operation signal output from the collision probability correction / calculation unit 20, the alarm device 31 generates an alarm sound generated from the speaker and an alarm to be displayed on the monitor by the speaker control device and the monitor control device, respectively. Each is generated on a speaker and displayed on a monitor.

  The collision avoidance assistance device 32 includes an automatic steering control device and an automatic steering device. The collision avoidance assisting device 32 is an automatic steering control device that is based on the collision response operation signal output from the collision probability correction / calculation unit 20 and is used in an automatic steering device for avoiding a collision or alleviating an impact caused by a collision. A steering direction and a steering amount are calculated. The collision avoidance assistance device 32 controls the automatic steering device based on the steering direction and the steering amount calculated by the automatic steering control device.

  The seat belt control device 33 controls the tightening force of the seat belt. The seat belt control device 33 calculates the tightening force of the seat belt based on the collision response operation signal output from the collision probability correction / calculation unit 20, and adds the tightening force to the seat belt as necessary.

  The seat position control device 34 controls a seat position moving device that automatically adjusts the seat position of the vehicle. The seat position control device 34 calculates a seat position for alleviating the impact on the occupant due to the collision based on the collision response operation signal output from the collision probability correction / calculation unit 20. The sheet position control device 34 controls the sheet position moving device based on the calculated sheet position.

  The brake control device 35 automatically controls the brake to give a vehicle deceleration force. The brake control device 35 calculates a deceleration force to be applied to the vehicle based on the collision response operation signal output from the collision probability correction / calculation unit 20, and controls the brake based on the calculated deceleration force.

  The airbag control device 36 controls the deployment of the airbag and its deployment preparation. The airbag control device 36 determines whether the airbag is to be deployed or prepared for deployment based on the collision response operation signal output from the collision probability correction / calculation unit 20, and based on the determination, the airbag control device 36 Control the bag.

  Subsequently, a control procedure of the collision prediction apparatus according to the present embodiment will be described. FIG. 2 is a flowchart showing a control procedure of the collision prediction apparatus according to the present embodiment.

  As shown in FIG. 2, the collision prediction ECU 1 first acquires opponent vehicle information based on the reflected wave information output from the millimeter wave radar sensor 5 (S1). Specifically, the opponent vehicle speed calculation unit 14 calculates the opponent vehicle speed, and the opponent vehicle movement distance calculation unit 15 calculates the opponent vehicle movement distance. Further, the opponent vehicle lateral position calculation unit 16 calculates the opponent vehicle lateral position.

  Next, an estimated curve radius is calculated (S2). The estimated curve radius is calculated by the estimated curve radius calculation unit 11 based on the steering angle information output from the steering angle sensor 2 and the yaw rate information output from the yaw rate sensor 3. After the calculation of the estimated curve radius, the vehicle speed is calculated (S3). The calculation of the host vehicle speed is performed by the host vehicle speed calculation unit 13 based on the wheel speed information output from the wheel speed sensor 4. After the calculation of the host vehicle speed, the host vehicle track line is calculated (S4). The calculation of the own vehicle track line is performed in the own vehicle track calculation unit 12 based on the estimated curve radius output from the estimated curve radius calculation unit 11.

  When the calculation of the own vehicle track line is completed, the opponent vehicle track line is calculated (S5). The opponent vehicle trajectory line is calculated in the opponent vehicle trajectory calculation unit 17 based on the opponent vehicle movement distance output from the opponent vehicle movement distance calculation unit 15 and the opponent vehicle lateral position output from the opponent vehicle lateral position calculation unit 16. The The opponent vehicle track line is performed using a plurality of opponent vehicle lateral positions. For this reason, when calculating the opponent vehicle track line, a plurality of opponent vehicle lateral positions are required. The opponent vehicle trajectory calculation unit 17 stores the opponent vehicle lateral position output from the opponent vehicle lateral position calculation unit 16, and starts calculating the opponent vehicle track line when a plurality of opponent vehicle lateral positions are arranged. When a plurality of opponent vehicle lateral positions are obtained and a plurality of opponent vehicle lateral positions are obtained, an opponent vehicle track line is calculated by performing linear regression from the plurality of opponent vehicle lateral positions using a least square method or the like.

  Subsequently, the collision probability calculation unit 18 calculates the collision probability between the host vehicle and the opponent vehicle (S6). In calculating the collision probability, first, the collision position calculation unit 21 outputs the own vehicle speed output from the own vehicle speed calculation unit 13, the opponent vehicle speed output from the opponent vehicle speed calculation unit 14, and further output from the own vehicle track calculation unit 12. The collision position of the opponent vehicle with respect to the width direction of the own vehicle is included in any of a plurality of divisions divided in advance based on the own vehicle track and the opponent vehicle track output from the opponent vehicle track calculation unit 17. Calculate. Next, a section including the collision position of the opponent vehicle is calculated through a similar process after a predetermined time, for example, 10 ms.

  Thereafter, similarly, a section including the collision position of the opponent vehicle is calculated, and a plurality of sections including the collision position of the opponent vehicle are obtained. Next, the collision position distribution calculation unit 22 calculates the collision position distribution by adding the collision position count value to the section including the collision position of the opponent vehicle obtained by the collision position calculation unit 21.

  Here, FIG. 3 shows an example of the collision position distribution. In the example shown in FIG. 3, 15 collision position count values are added to the host vehicle X, and among the first section D1 to the fifth section D5, three collision position meters are arranged in the second section D3. Numerical values are added, eight collision position count values of the third section D3 are added, and four collision position count values are added to the fourth section D4.

  After the collision position distribution is calculated in this way, the collision probability is calculated based on the collision position distribution. The collision probability calculated here is calculated as a numerical value that increases the collision probability in a section having a larger collision position count value.

  After the calculation of the collision probability, the collision probability calculation unit 18 calculates the partner vehicle lateral speed based on the host vehicle track, the partner vehicle track, and the partner vehicle speed, and outputs it to the collision probability correction / calculation unit 20. Thereafter, the collision probability correction / calculation unit 20 corrects the collision probability calculated by the collision probability calculation unit 18 (S7). When correcting the collision probability, the opponent vehicle lateral position output from the opponent vehicle lateral position calculator 16 and the opponent output from the image processor 19 are compared with the collision probability output from the collision probability calculator 18. A correction process is performed based on the relative positional relationship between the opponent vehicle lateral position and the opponent vehicle edge calculated from the vehicle edge and, further, the opponent vehicle lateral velocity output from the collision probability calculation unit 18. The correction process will be described later.

  When the correction process is completed, it is determined whether or not the own vehicle collides with the opponent vehicle (S8). This determination is made based on whether or not the collision position count value in each section exceeds a predetermined threshold value set in each section. However, this collision determination can be performed in various ways. Further, the collision position is also determined from the collision position count value.

  As a result, when it is determined that the host vehicle collides with the opponent vehicle, a collision response operation signal is transmitted to the collision avoidance device to activate the collision avoidance device (S9). Collision position information is attached to the collision response operation signal. Specifically, when the collision avoidance device is operated, an alarm is issued from the alarm device 31 that the vehicle collides with the section corresponding to the collision position information, or the seat belt on the side close to the section corresponding to the collision position information is tightened. The seat belt control device 33 is operated so that the force becomes stronger. And the process by a collision prediction apparatus is complete | finished. On the other hand, if it is determined in step S8 that the host vehicle does not collide with the opponent vehicle, the process by the collision prediction device is terminated as it is.

  Now, the correction process based on the relative positional relationship between the opponent vehicle lateral position and the opponent vehicle edge in step S7 will be described with reference to FIGS. 4 (a) to 4 (c) and FIGS. 5 (a) and 5 (b) show the positional relationship between the own vehicle and the partner vehicle and the detection results in the millimeter wave radar sensor 5 and the image sensor 6, respectively. It is a figure shown typically. FIGS. 6A to 6C are maps each showing a correction ratio when the collision probability is corrected.

  In performing the correction process, the opponent vehicle edge position that is the position of the opponent vehicle edge and the opponent vehicle edge width that is the length in the width direction are calculated. Next, the opponent vehicle lateral position is compared with the opponent vehicle edge, and the position of the opponent vehicle lateral position with respect to the opponent vehicle edge is calculated. The collision probability is corrected based on the position of the opponent vehicle lateral position with respect to the opponent vehicle edge and the opponent vehicle edge width. Specifically, the collision probability is corrected as follows.

  As shown in FIG. 4A, the opponent vehicle edge E detected by the image sensor 6 is detected in front of the host vehicle X, and the opponent vehicle lateral position P detected by the millimeter wave radar sensor 5 is the opponent vehicle edge. It is assumed that it is located at approximately the center in the width direction of E. In this case, it is considered that the opponent vehicle Y can be accurately detected by the millimeter wave radar sensor 5 and the image sensor 6. Further, it is considered that the detection accuracy by the image sensor 6 is higher as the opponent vehicle lateral position P is closer to the center of the opponent vehicle edge E. Therefore, since it is considered that the collision probability of the opponent vehicle Y with respect to the own vehicle X is increased, the correction is performed so that the collision probability is increased.

  Here, when correcting the collision probability, a map shown in FIG. 6A can be used. The map shown in FIG. 6A shows the relationship between the position in the width direction of the opponent vehicle edge E at the opponent vehicle lateral position and the collision probability correction value. In this map, the collision probability increase range is increased in the range from the center of the opponent vehicle edge to the vicinity of the opponent vehicle edge. In a part of the range on the outer left and right sides of this range, the collision probability increase width is defined so that the increase width gradually decreases as it moves to the end. Further, in the range on the left and right sides of this range, the collision probability increase width is set to a predetermined reference value. By defining the collision probability increase range in this way, a highly accurate collision probability can be calculated.

  In addition, as shown in FIG. 4B, in the detection by the image sensor 6, in addition to the opponent vehicle Y, the opponent vehicle Y and the object Z including the object Z between the opponent vehicle Y and the own vehicle X are included. May be erroneously detected as an integrated object. In this example of erroneous detection, the image sensor 6 detects that the opponent vehicle Y and the object Z are an integrated object, and the millimeter wave radar sensor 5 erroneously detects the object Z as the opponent vehicle Y.

  When the partner vehicle Y and the object Z are erroneously detected as an integrated object, the partner vehicle lateral position P is detected at the end of the partner vehicle edge E. As an example of such a situation, there is, for example, a situation in which a building or a sign located on the side of the road is judged as an integrated object with the opponent vehicle while the host vehicle is traveling on a curve.

  In this case, the opponent vehicle edge E occurs at a position different from that of the opponent vehicle Y. At this time, it is determined that the collision probability is high even though there is still a distance until the vehicle actually collides with the opponent vehicle Y. As a result, although the collision probability with the opponent vehicle Y is not actually high, the collision avoidance device is activated.

  Therefore, when there is a possibility of erroneous detection as shown in FIG. 4B, the collision probability is corrected to be lowered. In this way, when the opponent vehicle lateral position P is detected at the end of the opponent vehicle edge E, the collision avoidance device is prevented from operating early by correcting the collision probability to be low. Can do.

  Further, as shown in FIG. 4C, the objects Z1 and Z2 on both sides of the front side of the host vehicle X may be erroneously detected as the partner vehicle Y when detected by the image sensor 6. In this erroneous detection example, the image sensor 6 erroneously detects the objects Z1 and Z2 as the counterpart vehicle, and the millimeter wave radar sensor 5 erroneously detects the first object Z1 as the counterpart vehicle.

  When the objects Z1 and Z2 on both front sides of the host vehicle X are erroneously detected as the partner vehicle, the partner vehicle edge E is generated between the objects Z1 and Z2, although the partner vehicle does not actually exist. In this case, there is actually no opponent vehicle Y to be a collision target. Further, regarding the relationship with the objects Z1 and Z2, even though the host vehicle X can pass between the objects Z1 and Z2, it is determined that the collision probability is high. As a result, although the collision probability with the opponent vehicle Y is not actually high, the collision avoidance device is activated.

  Therefore, when there is a possibility of erroneous detection as shown in FIG. 4C, correction is performed so as to reduce the collision probability. In this way, by correcting the collision probability of the objects Z1 and Z2 on both sides ahead of the host vehicle X so that the collision probability is low when the vehicle is erroneously detected, the early operation of the collision avoidance device can be prevented.

  Furthermore, when the opponent vehicle lateral position P is close to the center of the opponent vehicle edge E as in the example of FIG. 4A, the edge accuracy, which is the accuracy to which the same object belongs, is increased. On the other hand, as shown in the examples of FIGS. 4B and 4C, when the opponent vehicle lateral position P is far from the center of the opponent vehicle edge E (close to the end), the plurality of edges are the same. The edge accuracy, which is the accuracy to which the object belongs, is lowered. Therefore, for example, the distance from the center of the opponent vehicle edge E to the opponent vehicle lateral position P can be used as the edge accuracy, and the collision probability can be corrected according to the edge accuracy. According to such a correction mode, when calculating the collision probability of the opponent vehicle Y with respect to the host vehicle X, it is possible to calculate a highly accurate collision probability.

  In addition, as shown in FIG. 5A, the opponent vehicle Y may be located in front of the host vehicle X. In this case, the opponent vehicle Y may have an offset collision with the host vehicle X. May be expensive. In the example shown in FIG. 5 (a), the opponent vehicle Y is located on the left front side of the host vehicle X. The opponent vehicle edge E detected by the image sensor 6 has an end portion on the side close to the host vehicle X (hereinafter, referred to as the partner vehicle Y). There is a “near end” and a far end (hereinafter “far end”). Further, the opponent vehicle lateral position P is located on the opponent vehicle edge E, and it is considered that the detection by the millimeter wave radar sensor 5 and the image sensor 6 was normal.

  In this case, when the opponent vehicle lateral position P is located in the vicinity of the near-side end portion of the opponent vehicle edge E, it is considered that the possibility of a collision is higher than that in the case of being located at the far-side end portion of the opponent vehicle edge E. Further, when the opponent vehicle lateral position P is located in the vicinity of the far end of the opponent vehicle edge E, it is considered that the opponent vehicle Y is more likely to travel in a position where it does not collide with the own vehicle X. It is done. Therefore, in the present embodiment, when the opponent vehicle lateral position P is located in the vicinity of the near end of the opponent vehicle edge E, the collision probability is higher than when it is located at the far end of the opponent vehicle edge E. The collision probability is corrected.

  Further, as shown in FIG. 4B, when the object Z is present on the front side of the host vehicle X, the object Z is erroneously detected as the partner vehicle Y, and as shown in FIG. It is conceivable that E is detected. However, in this case, since the millimeter wave radar sensor 5 detects the object Z, the opponent vehicle lateral position P may be detected at a position other than the vicinity of the near end of the opponent vehicle edge E. Conceivable. For this reason, when the opponent vehicle lateral position P is located in the vicinity of the far end of the opponent vehicle edge E, the collision probability is corrected so that the collision probability increase width is smaller than a predetermined reference value.

  Specifically, for example, a map shown in FIG. 6B can be used as a map for correcting the collision probability. In the map shown in FIG. 6B, the collision probability increase range is increased in the range from the center in the width direction of the opponent vehicle edge E to the vicinity thereof. Further, in a partial range on the near side end portion side and the far side end portion side than this range, the collision probability increase width is defined so that the increase width gradually decreases as the position shifts to the end portion. Further, in the range closer to the outer end than this range, the collision probability increase width is set to a predetermined reference value.

  Further, the far end end side is wider in the range where the collision probability increase range is gradually smaller than the near end side than the range in which the collision probability increase range is increased. Further, in a part of the range on the far end side, the collision probability increase width is set to a value smaller than a predetermined reference value. By defining the collision probability increase range in this way, it is possible to calculate a collision probability with higher accuracy. In addition, when there is a possibility of causing an offset collision, it is possible to prevent the collision probability from being estimated too low.

  Further, as shown in FIG. 5B, the opponent vehicle Y may be located in front of the host vehicle X, and the opponent vehicle Y has a lateral velocity component directed toward the host vehicle X. There may be. In this case, in the example shown in FIG. 5B, the opponent vehicle Y approaches from the left direction of the host vehicle X with a lateral speed, and the opponent vehicle edge E includes a near end and a far end. There is. Here, when the opponent vehicle lateral position P is located at the far end of the opponent vehicle edge E, the opponent vehicle Y approaches the host vehicle X with a lateral velocity component as shown in FIG. In some cases, the stationary object Z on the side of the host vehicle X is erroneously detected as the opponent vehicle Y, as shown in FIG. At this time, if the partner vehicle lateral position P corresponding to the detected partner vehicle Y has a lateral speed component, it is the partner vehicle Y, and the partner vehicle lateral position P does not have a lateral speed component. In this case, it is considered that the object Z is erroneously detected as the opponent vehicle Y.

Therefore, when the opponent vehicle lateral position P is located in the vicinity of the far side end of the opponent vehicle edge E, the opponent vehicle lateral position P does not have a lateral velocity component if it has a lateral velocity component. The collision probability is corrected so that the collision probability is higher than the case. Specifically, when correcting the collision probability, if the opponent vehicle lateral position P does not have a lateral velocity component and the map shown in FIG. When it has a component, the map shown in FIG. 6C is referred to.

  In the map shown in FIG. 6C, the collision probability increase range is increased in the range from the center in the width direction of the opponent vehicle edge E to the vicinity of the opponent vehicle edge E. Further, in a partial range on the near side end portion side and the far side end portion side than this range, the collision probability increase width is defined so that the increase width gradually decreases as the position shifts to the end portion. Further, in the range closer to the outer end than this range, the collision probability increase width is set to a predetermined reference value.

  Further, the far side end portion side is narrower in the range where the collision probability increase width is gradually smaller than the near side end portion side than the range in which the collision probability increase width is increased. Furthermore, in a part of the range on the far side end side, the collision probability increase width is set to a value larger than a predetermined reference value. By defining the collision probability increase range in this way, it is possible to prevent the collision probability of the opponent vehicle Y against the own vehicle X from being excessively low when the opponent vehicle lateral position P is located at the far end of the opponent vehicle edge E. be able to. And the collision probability of the other vehicle Y with respect to the own vehicle X can be determined still more accurately.

  The preferred embodiment of the present invention has been described above, but the present invention is not limited to the above embodiment. For example, in the above embodiment, a millimeter wave radar sensor is used as the radar sensor, but other radar sensors such as a laser radar sensor can be used. In the above-described embodiment, the collision position is also determined when the collision position is determined. However, it is also possible to simply perform the collision determination.

It is a block block diagram which shows the structure of the collision prediction apparatus which concerns on this invention. It is a flowchart which shows the control procedure of a collision prediction apparatus. It is a typical top view which shows the collision position distribution in the own vehicle. (A)-(c) is a figure which shows typically the positional relationship of the own vehicle and the other party vehicle, and the detection result in each of a millimeter wave radar sensor and an image sensor. (A), (b) is a figure which shows typically the positional relationship of the own vehicle and the other party vehicle, and the detection result in each of a millimeter wave radar sensor and an image sensor. (A)-(c) is a map which shows the correction | amendment ratio at the time of performing correction | amendment of a collision probability all.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Collision prediction ECU, 2 ... Steering angle sensor, 3 ... Yaw rate sensor, 4 ... Wheel speed sensor, 5 ... Millimeter wave radar sensor, 6 ... Image sensor, 11 ... Estimated curve radius calculation part, 12 ... Own vehicle track | orbit calculation part , 13 ... Own vehicle speed calculation unit, 14 ... Counter vehicle speed calculation unit, 15 ... Counter vehicle movement distance calculation unit, 16 ... Counter vehicle lateral position calculation unit, 17 ... Counter vehicle trajectory calculation unit, 18 ... Collision probability calculation unit, 19 ... Image processing unit, 20 ... collision probability correction / calculation unit, 21 ... collision position calculation unit, 22 ... collision position distribution calculation unit, 31 ... alarm device, 32 ... collision avoidance support device, 33 ... seat belt control device, 34 ... seat Position control device, 35 ... Brake control device, 36 ... Airbag control device, E ... Counter vehicle edge, P ... Counter vehicle lateral position, X ... Host vehicle, Y ... Counter vehicle, Z (Z1, Z2) ... Object.

Claims (5)

Radar detecting means for detecting a target object by radar;
An image detecting means for detecting an edge from an image picked up by the image pickup means and detecting the target object based on the detected edge;
A collision probability calculating means for calculating a collision probability between the target object and the host vehicle based on the relationship between the target object and the host vehicle;
A radar lateral position that is a position in the vehicle width direction of the target object detected by the radar detection unit and an edge width at a target object corresponding edge that is an edge corresponding to the target object detected by the image detection unit A relative positional relationship detecting means for detecting a relative positional relationship with the direction position;
Collision probability correction means for correcting the collision probability between the target object and the host vehicle based on the relative positional relationship between the radar lateral position and the edge width direction position at the target object corresponding edge;
A collision prediction apparatus comprising: The collision probability correcting means includes
When the radar lateral position is located at the edge of the target object corresponding edge in the edge width direction, the collision probability is relatively smaller than when the radar is positioned at the center of the target object corresponding edge in the edge width direction. The collision prediction apparatus according to claim 1, wherein the collision prediction apparatus corrects to be low. The collision probability correcting means includes
When the radar lateral position is located at an end of the target object-corresponding edge on the side close to the host vehicle in the edge width direction, an end of the target object-corresponding edge on the side far from the host vehicle in the edge width direction The collision prediction apparatus according to claim 1, wherein the collision prediction apparatus corrects the collision probability so as to be relatively higher than a case where the collision probability is relatively high. A lateral speed detection means for detecting whether the target object has a lateral speed component that is a speed component along a direction orthogonal to the traveling direction of the host vehicle and detects whether the target object is approaching the host vehicle;
The collision probability correcting means is more suitable for the target object when the target object is approaching with the lateral velocity component than when the target object is not approaching with the lateral velocity component. 4. The collision probability according to claim 1, wherein the collision probability is corrected so that the collision probability is relatively high when the edge is located at an end of the edge in the edge width direction far from the host vehicle. The collision prediction apparatus according to item. Radar detecting means for detecting a target object by radar;
Image detecting means for detecting a plurality of edges from an image captured by the imaging means and detecting the target object based on the detected plurality of edges;
A collision probability calculating means for calculating a collision probability between the target object and the host vehicle based on a relationship between the target object detected by the radar detecting unit and the image detecting unit and the host vehicle;
Edge accuracy detection means for detecting edge accuracy, which is the accuracy that the plurality of edges belong to the same object;
Collision probability correcting means for correcting the collision probability according to the edge accuracy;
A collision prediction apparatus comprising:

Images