Device for calculating the probability of a collision for a vehicle and a collision avoidance system using this device.

Abstract

Device (10) for calculating the probability of a collision for a vehicle, comprising: an object information acquisition device (S110) for acquiring object information including the position of an object in the vicinity of the vehicle, wherein the object is referred to as a nearby object; a recording device (S120) for vehicle condition information to record information about the vehicle condition including the vehicle's speed and direction of travel; a collision probability map generating device (S310) for generating a collision probability map based on the acquired information about the vehicle state, which indicates a collision probability that the own vehicle will collide with a nearby object at any point in an area in the vicinity of the own vehicle; a traffic lane width detection device (S410) for detecting the width of a traffic lane in which the vehicle is moving, wherein this traffic lane is referred to as a driving lane; a collision probability distribution correction tool (S420) for correcting a collision probability distribution in the collision probability map in order to produce a corrected collision probability map in response to the detected lane width; a collision probability correction tool (S570, S580) for reading a collision probability associated with the position of the nearby object from the corrected collision probability map and correcting a collision probability distribution in the collision probability map to generate a corrected collision probability map and to correct the read collision probability in response to the captured object information; a collision probability output means (S590) to output the collision probability corrected by the collision probability correction means (S570, S580) as a more accurate collision probability that the own vehicle will collide with the nearby object; a recording image acquisition device (S110) for capturing a recorded image of the surroundings of the vehicle; a lane edge detection device (S320) for detecting the edge of the lane of one's own vehicle in the recorded image; an area extraction agent (S520) for extracting an area surrounding and in the vicinity of the nearby object from the captured image and a lane edge detection device (530) for determining the presence of a lane edge to determine whether a lane edge of the own vehicle is present in the extraction area, wherein the collision probability correction means (S570, S580) corrects the read collision probability to a higher value if the lane edge detection means (S530) detects that an edge of the lane of the own vehicle is present in the extraction area near the nearby object; a motion velocity detection device (S510) for detecting a motion velocity of the nearby object, wherein the motion velocity of the nearby object is included in the object information of the nearby object, wherein the area extraction agent (S520) is suitable to change the dimensions of an area to be extracted from the recorded image (referred to as the extraction area) as a function of the speed of movement of the nearby object.

Description

BACKGROUND (Technical Area)

[0001] This invention relates to a collision probability calculating device for calculating a probability that the own vehicle will collide with an object in the vicinity of the own vehicle, and a collision avoidance system using the same. (State of the art)

[0002] JP 2008-27309 A discloses the following solution: Image data acquired by a plurality of imaging devices imaging the surroundings of a vehicle is acquired, the presence of an obstacle in an image is detected, and the relative position data of the obstacle to the vehicle is detected time-sequentially, so that the possibility of a collision can be determined. The peripheral area of ​​the vehicle is divided into several areas based on its relative position to the vehicle, and a parameter indicating the degree of collision risk is stored in each of the divided areas. The corresponding area is extracted based on the acquired relative position data, and the parameter corresponding to the extracted area is read, so that the possibility of a collision is determined.

[0003] JP 2009-271766 A discloses the following solution: Obstacle detection areas are set according to a predicted route of the vehicles and a specific road structure, such as a pedestrian crossing or a narrow road, around the vehicle to detect obstacles in the obstacle detection areas. Thus, it is possible to set the obstacle detection areas according to a pedestrian crossing or a narrow road where the vehicle is highly likely to collide with the obstacles. Therefore, it is possible to reduce the blind spot of a road and improve obstacle detection performance.

[0004] JP 2010 - 176 591 A discloses the following solution: A secondary warning zone extends 50 meters ahead, and when a pedestrian is detected within the secondary warning zone, a red LED element generates secondary warning information. On this side, within a radius of 50 meters, there is a primary warning zone that occupies areas outside the right and left sides that extend beyond the right and left white lines. When an obstacle is detected within the primary warning zone, a green LED element of an LED display device generates primary warning information. For the primary warning information and the secondary warning information, the LED elements are controlled so that a warning mark A is reflected on a windshield in a section corresponding to the proximity of the pedestrian's feet when the driver views the pedestrian from an eyepoint on a virtual line connecting the pedestrian and the eyepoint.

[0005] US 2007 / 0 069 873 A1 discloses a vehicle environment monitoring system comprising a vehicle travel information acquisition unit for acquiring travel information of a self-vehicle, a solid-state detection unit for detecting solid-state objects present in front of the self-vehicle and detecting at least one preceding vehicle from the solid-state objects, a target route calculation unit for calculating a target route for the self-vehicle according to the current position of the preceding vehicle and the current position of the self-vehicle, an alarm area setting unit for setting an alarm area based on the target route, and an alarm control unit for triggering an alarm according to a current state of solid-state objects present in the alarm area.

[0006] A known device for calculating the collision probability for a vehicle, as disclosed in Japanese Patent No. JP 4 026 400 B2, operates in such a way that future positions of the own vehicle and a target object in the vicinity of the own vehicle, such as a pedestrian, are estimated on the basis of a current behavior of the own vehicle and a current position of the target object, and based on this estimation, a probability is calculated that the own vehicle will collide with the target object.

[0007] However, according to the techniques described in Japanese Patent No. JP 4 026 400 B2, despite the behavior of the own vehicle changing over time, the device predicts the future position of the own vehicle based on the current behavior, which may result in low accuracy in predicting the future position of the own vehicle. Therefore, the disclosed device cannot accurately predict the collision probability.

[0008] In view of the foregoing, it is an object of the present invention to provide a collision probability calculation device capable of more accurately calculating the probability that the own vehicle collides with an object in the vicinity of the own vehicle. SUMMARY OF THE INVENTION

[0009] The object is achieved by the device having the features of claim 1. Further advantageous developments and embodiments of the invention are the subject of the subsequent claims.

[0010] According to one embodiment of the present invention, there is provided a collision probability calculation device for a vehicle, comprising: object information acquiring means for acquiring object information including a position of an object within the vicinity of the own vehicle (referred to as a nearby object); vehicle state information acquiring means for acquiring vehicle state information including a traveling speed and a traveling direction of the own vehicle; and collision probability map generating means for generating a collision probability map based on the acquired vehicle state information, indicating a collision probability that the own vehicle will collide with a nearby object at each point of an area within the vicinity of the own vehicle.

[0011] The apparatus further includes means for acquiring the width of a traffic lane in which the host vehicle is traveling (referred to as a lane), means for correcting a collision probability distribution in the collision probability map to generate a corrected collision probability map in response to the acquired lane width;correction means for reading out a collision probability associated with the position of the nearby object from the corrected collision probability map and correcting a collision probability distribution in the collision probability map to generate a corrected collision probability map and correcting the read collision probability in response to the detected object information; and output means for outputting the collision probability corrected by the correction means as a more accurate collision probability that the host vehicle will collide with the nearby object;

[0012] This allows the collision probability map (i.e., the collision probability distribution) to be corrected in response to lane width, so that a collision probability that the host vehicle will collide with a nearby object located in an area where the host vehicle is more likely to move is higher than a collision probability that the host vehicle will collide with a nearby object located in an area where the host vehicle is less likely to move. Therefore, an area where the host vehicle is likely to move can be more accurately detected, leading to a more accurate calculation of the collision probability.

[0013] According to one embodiment, the collision probability distribution correcting means sets a width of a range in which the collision probability is constant (referred to as a constant collision probability range) such that the width of the constant collision probability range decreases with a decreasing width of the lane.

[0014] It can be assumed that the host vehicle is more likely to stay in the lane without deviating from it as the lane width decreases. Therefore, it is preferable to reduce the width of the constant collision probability area in the collision probability map, making it less likely that the host vehicle will move outside the lane. This allows the collision probability in the collision probability map to be more accurately corrected in response to the lane width.

[0015] According to one embodiment, the device further includes: captured image acquisition means for acquiring a captured image of the surroundings of the own vehicle; lane edge detection means for detecting an edge of the lane of the own vehicle in the captured image; surrounding area extraction means for extractingExtracting an area surrounding and in the vicinity of the nearby object from the captured image and a lane edge presence detection means for determining whether a lane edge of the own vehicle is present in the extraction area, wherein the collision probability correction means corrects the extracted collision probability to a higher value when the lane edge presence detection means determines that an edge of the own vehicle's lane is present in the extraction area in the vicinity of the nearby object.

[0016] That is, when a lane edge is within the sampling range, the collision probability read from the collision probability map is adjusted to a higher value in response to the lane width. This allows the collision probability of the host vehicle colliding with a nearby object that is moving or within the host vehicle's movement range to be set higher, because such a nearby object can be considered a more dangerous nearby object (an object with which the host vehicle is more likely to collide).

[0017] According to one embodiment, the device further includes a detection means for acquiring a movement speed of the nearby object. The movement speed of the nearby object is included in the object information of the nearby object. The surrounding area extraction means changes the dimensions of an area to be extracted from the captured image (referred to as extraction area) as a function of the movement speed of the nearby object.

[0018] By this design, the dimensions of the sampling area are changed as a function of the moving speed, so that the probability of a lane edge being present in the sampling area can be changed as a function of the moving speed of the nearby object, which allows to facilitate the change of the collision probability in response to the moving speed of the nearby object.

[0019] According to one embodiment, the apparatus further includes vehicle control means for controlling the own vehicle to avoid a possible collision with the nearby object in response to the collision probability outputted by the collision probability output means.

[0020] This design allows the driver to control the vehicle based on a precisely calculated collision probability, ensuring that only necessary vehicle control can be performed.

[0021] According to one embodiment, the apparatus further includes delay control means for delaying the start of the collision avoidance control in response to the collision probability outputted by the output means.

[0022] This design makes it less likely that a vehicle control considered unnecessary will be executed because the start of the vehicle control is delayed when the collision probability is below a predetermined threshold.

[0023] According to one embodiment, the device further includes a vulnerable road user detection means for determining whether or not the nearby object is a vulnerable road user likely to be injured by the host vehicle. The collision probability correction means corrects the read collision probability to a first higher value if the lane edge presence detection means detects that an edge of the host vehicle's lane is present in the sampling area near the nearby object and the vulnerable road user detection means detects that the nearby object is a vulnerable road user likely to be injured by the host vehicle, and corrects the read collision probability to a second higher value.which is lower than the first higher value if the means for detecting the presence of a lane edge detects that an edge of the lane of the own vehicle is present in the sampling area near the nearby object and the means (S560) for detecting a vulnerable road user detects that the nearby object is not a vulnerable road user likely to be injured by the own vehicle.

[0024] By this design, the corrected collision probability is output in such a way that the driver's attention is directed to the vulnerable road user because the collision probability read from the collision probability map is corrected to a higher value when the nearby object is a vulnerable road user who is likely to be injured by the own vehicle.

[0025] According to one embodiment, there is provided a collision avoidance system for a vehicle, including: a collision probability calculation device according to any one of the embodiments of the present invention, in which the vehicle control means controls the host vehicle by issuing a collision avoidance control command in response to the collision probability outputted by the collision probability outputting means.

[0026] The system further includes a near object detector for a nearby object that detects at least a position of an object (referred to as a nearby object) within the vicinity of the own vehicle and outputs the detected position of the nearby object to the device; a vehicle state detector for the vehicle state that detects at least a traveling speed and a traveling direction of the own vehicle and outputs the detected traveling speed and traveling direction of the own vehicle to the device; and a collision avoidance control executor that executes collision avoidance control upon receiving a collision avoidance control command from the device.

[0027] In this way, a system can be established that acts as a collision avoidance system for a vehicle according to an embodiment of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The attached drawings show Fig. 1A schematically shows a block diagram of a collision avoidance system according to an embodiment of the present invention Fig. 1B one in a RAM of the Fig. 1A stored object information for each target object; Fig. 2 a flowchart of a collision avoidance procedure; Fig. 3A is a flowchart of a method for calculating the collision probability of the Fig. 2; Fig. 3B is a flowchart of a method for generating a collision probability map of the Fig. 3A; Fig. 4A an example of a collision probability base map; Fig. 4B shows an example of a corrected collision probability map; Fig. 5A is a flowchart of a method for correcting a map of the Fig. 3B; Fig. 5B is a flowchart of a method for correcting a collision probability of the Fig. 3B; Fig. 6 is an exemplary diagram comparing the correction factor with the maximum width of an area for a collision probability of 80%; Fig. 7A is an exemplary diagram comparing the width of the evaluation area with the movement speed of a pedestrian; Fig. 7B an example of an evaluated area; Fig. 8 an example of a method for correcting the collision probability of the Fig. 5B; Fig. 9 a flowchart of an activation decision procedure of the Fig. 2 and Fig. 10 is an exemplary diagram comparing the correction value of the activation time with the collision probability. DESCRIPTION OF SPECIAL EMBODIMENTS

[0029] The present invention will be described more fully below with reference to the accompanying drawings. Like reference numerals refer to like elements throughout.

[0030] The Fig. 1A schematically illustrates a block diagram of a collision avoidance system 1 according to an embodiment of the present invention. The collision avoidance system 1, which is installed in a vehicle (hereinafter referred to as the own vehicle), such as a passenger car, determines whether or not there is a risk of the own vehicle colliding with an object in the vicinity of and outside the own vehicle (hereinafter also referred to as a nearby object). If it is determined that there is a risk of the own vehicle colliding with the nearby object, the collision avoidance system 1 performs collision avoidance control and / or triggers an alarm. Hereinafter, the nearby object detected by image processing or the like is referred to as a target object.

[0031] As in Fig. As shown in FIG. 1A, the collision avoidance system 1 includes a microcomputer 10 (a collision probability calculation device), a near object detector 21 for a nearby object, a vehicle state detector 22 for the vehicle state, a collision avoidance control execution means 24, and an alarm indicator 25. Alternatively, the collision avoidance system 1 may be configured to have either the collision avoidance control execution means 24 or the alarm indicator 25.

[0032] The microcomputer 10, as is well known, includes a CPU 11, a ROM 13, a RAM 15, and others. The microcomputer 10 (CPU 11) executes various processes, including a collision avoidance process (described later), through programs stored in the ROM 13, programs loaded in the RAM 15, and detection results of the near-object detector 21 and the vehicle state detector 22.

[0033] The near object detector 21, which is equipped with a camera, records the surroundings of the own vehicle (mainly in the forward direction of the own vehicle) and outputs the recorded images to the microcomputer 10. The microcomputer 10 receives the recorded images, detects edge points (such that each edge point is located between adjacent pixels that have a greater luminance difference than a predetermined value) in the recorded image, and stores them as shown in Fig. 1B shows the positions of the edge points in a memory such as RAM 15.

[0034] The near-object detector 21 includes, but is not limited to, a well-known radar device or a well-known sonar device, or both. The radar device or the sonar device detects a relative displacement vector (a position, a speed, a direction) of the nearby object relative to the host vehicle by transmitting an electromagnetic wave (such as a light wave and a radio wave) or a sound wave and receiving the wave reflected from the object close to the host vehicle, and outputs the detected relative displacement vector to the microcomputer 10. The microcomputer 10 stores, as shown in Fig. 1B, the vector of relative displacement in a memory such as the RAM 15,

[0035] The positions of the edge points and the relative displacement vector are included in the object information.

[0036] The vehicle state detector 22 includes, but is not limited to, a well-known speed sensor and a well-known steering angle sensor, and outputs vehicle state information including at least a traveling speed and a steering angle (traveling direction) of the host vehicle to the microcomputer 10.

[0037] The collision avoidance control execution part 24 controls the hydraulic braking pressure and the tensioning force of a seat belt of the host vehicle. Upon receiving a collision avoidance control instruction from the microcomputer 10, the collision avoidance control execution part 24 stops the movement of the host vehicle or fastens the seat belt of the host vehicle driver to reduce or prevent injuries caused by the collision impact.

[0038] The alarm indicator 25 is equipped with a display (not shown) or a speaker (not shown), or both. Upon receiving an alarm command from the microcomputer 10, the alarm indicator 25 warns the driver of a potential collision by displaying the alarm on the display or outputting a sound (including voice) through the speaker. (Collision avoidance procedure)

[0039] It will now be discussed with reference to the Fig. 2 explains a collision avoidance method that is carried out in the collision avoidance system 1. The Fig. 2 shows a flowchart of the collision avoidance process executed by the CPU 11 in the microcomputer 10.

[0040] The collision avoidance process is started, for example, with power on the microcomputer 10, whereupon the process is repeated periodically, for example, with a time interval of 100 msec.

[0041] First, in the collision avoidance method, the object information is acquired from the RAM 15 (object information requesting means, captured image requesting means) at step S110, and the vehicle state information is acquired from the vehicle state detector 22 (vehicle state information requesting means) at step S120.

[0042] In the step of acquiring the object information, the positions of the edge points and the vectors of the relative displacement (which are the detection results of the near-object detector 21 for a nearby object) stored in the RAM 15 are read out therefrom (see also Fig. 1B). In the vehicle state information acquisition step, the vehicle state detector 22 detects the vehicle state information, the vehicle speed and steering angle of the host vehicle (the detection result of the vehicle state detector 22).

[0043] Subsequently, in step S130, as shown in Fig. 1B, a complement of the target object is determined and stored in a memory such as RAM 15. The complement of the target object is included in the object information. The complement of the target object (e.g., whether the target object is a vehicle, a pedestrian, a bicycle, a motorcycle, a telephone pole, or something else) is explored using image processing such as well-known pattern matching or the like based on the positions of the edge points. In addition, a position of the target object is determined based on a position and a size of the target object in the captured image.

[0044] However, in the presence of the relative displacement vectors stored in RAM 15, the position of the target object is determined by one of the. Furthermore, in the target object complement detection step, whether the target object is a stationary object or a moving object is determined by the target object's movement speed. The movement speed of the target object can be determined based on the relative displacement vector and the vehicle state information.

[0045] Subsequently, in step S140, a collision risk is calculated. The collision risk is defined here by the collision time at which the host vehicle will collide with the target object. The collision time is calculated using the relative displacement vector or the driving speed of the host vehicle, as well as a relative distance (the distance from the host vehicle to the target object), among other factors. The collision risk increases as the collision time decreases.

[0046] Next, a collision probability calculation process is performed in step S150, and an activation decision process is performed in step S160. A control signal based on the activation decision process is output to the collision avoidance control execution unit 24 or the alarm indicator 25, or both, in step S170 (vehicle control means), whereupon the collision avoidance process is terminated. The host vehicle is thus controlled by the control signal output in response to the collision probability.

[0047] The procedure for calculating the collision probability is now explained. Fig. Figure 3A shows a flowchart of the process for calculating the collision probability within the collision avoidance process. Fig. Figure 3B is a flowchart of the method for generating a collision probability map within the method for calculating the collision probability. Fig. Figure 5A shows a flowchart of the map correction process within the collision probability map generation process. Fig. 5B shows a flowchart of the collision probability correction process within the collision probability map generation process.

[0048] The procedure for calculating the collision probability is as described in Fig. 3A, at step S210, the collision probability map generation process is performed, wherein a collision probability map indicating the probability of collision of the own vehicle with a nearby object is generated, and a collision probability at a position of a detected nearby object is outputted after some correction processes.

[0049] In particular, as in Fig. 3B, at step S310 (collision probability map generation means), a base collision probability map (hereinafter referred to as a base map) is generated based on the traveling speed and traveling direction of the own vehicle. This base map indicates a probability of collision of the own vehicle with a nearby object at each point within a range around the own vehicle. That is, the base map indicates a collision probability distribution determined by the vehicle state detector 22 and defined by the traveling speed and traveling direction of the own vehicle.

[0050] In the base map, as in Fig. As shown in Figure 4A, a frontal area in the forward direction of the host vehicle, where the host vehicle will collide with a nearby object at any point within the area within two seconds, is an area with a collision probability of 100%. The maximum width of the area with a collision probability of 100% is set equal to the width of the host vehicle. An area just outside and surrounding the area with a collision probability of 100%, where the host vehicle will collide with a nearby object at any point within the area within three seconds, is an area with a collision probability of 80%.

[0051] Since a driver typically tends to drive in the center of a lane (the lane in which the driver's own vehicle moves) while avoiding the left or right edges of the lane, the maximum width of the 80% collision probability area is 3.0 m, which is slightly smaller than the typical lane width of 3.5 m. An area just outside and surrounding the 80% collision probability area, where the driver's own vehicle will collide with a nearby object at any point in the area within five seconds, is a 40% collision probability area. An area where the probability that the driver's own vehicle will collide with a nearby object at any point in the area is extremely low is a 0% collision probability area.An area just outside the area with a 40% collision probability, but surrounded by the area with a 0% collision probability, is an area with a 20% collision probability.

[0052] Subsequently, by image processing such as the well-known baseline Hough transform of the captured image (an edge point distribution) at step S320 (lane edge detection means), a lane marker line of the own vehicle's lane (such as a white or yellow center line, lane line, or edge marker line, where the center marker line indicates the separation of opposing lanes, the edge marker line indicates the edge of a road, and the lane line indicates the separation of co-directional lanes) is detected. The map correction process is executed at step S330, wherein a collision probability distribution in the base map is corrected in response to the lane width.

[0053] The procedure for correcting the map is as described in Fig. 5A, a lane width is determined based on the positions of the marker lines detected in step S410 (lane width acquisition means). Subsequently, the collision probability base map is corrected in step S420 (collision probability distribution correction means) in response to the determined lane width. The collision probability base map is corrected, for example, by using a Fig. 6, ie a diagram comparing the correction factor with the lane width for a maximum width of the collision probability area of ​​80%.

[0054] Referring to the Fig. In the diagram shown in Figure 6, the correction factor is set to 1.0 when the lane width is 3.5 m. It is set to 1.1 when the lane width is equal to or greater than 3.8 m. The correction factor is set to 0.8 when the lane width is equal to or greater than 2.8 m and less than 3.8 m. For example, the width of the 80% collision probability range is set to 3.0 × 0.8 = 2.4 [m] when the lane width is 2.8 m.

[0055] For each of the other collision probability ranges, ie the 20%, 40% or 100% collision probability range, one of the Fig. 6 similar diagram can be used to correct the base map of collision probability.

[0056] As in the Fig. As can be seen from the corrected collision probability map shown in Figure 4B, the maximum width of the 80% collision probability range is set slightly smaller than the lane width.

[0057] After all ranges of collision probability are corrected in response to the lane width determined in step S420, the map correction process is terminated.

[0058] Back to the Fig. 3B, the collision probability correction process is then executed at step S340, wherein, if a marker line, such as a white or yellow center line, lane line, or edge marker line, is present near the target object and the target object is a vulnerable road user who could be injured by the host vehicle, the collision probability corrected at step S330 is further corrected to a higher value.

[0059] In particular, as in Fig. 5B, a moving speed of the target object is detected at step S510 (moving speed detecting means), where the moving speed of the target object is estimated based on the current and previously captured images. However, in the presence of the relative displacement vector stored in RAM 15, the moving speed is determined by one of the relative displacement vectors.

[0060] Thereafter, a region surrounding the target object centered on the target object is extracted from the captured image in step S520 (means for extracting the surrounding region) in response to the moving speed of the target object. For example, as shown in Fig. 7A, when the target object is a pedestrian, a width of an extraction area (an area to be extracted from the captured image) increases with an increasing speed of the pedestrian.

[0061] In Fig. 7A, when the pedestrian is stationary, the width of the withdrawal area is set to 1.2 m (“X1” in Fig. 7B). The width of the sampling area increases with the increasing pedestrian speed. If the pedestrian speed is 4 km / h, the width of the sampling area is set to 1.5 m. For a pedestrian speed of 8 km / h or greater, the width of the sampling area is set to 2.0 m (“X2” in Fig. 7B).

[0062] A height of the sampling area (“Y” in Fig. 7B) is a fixed value, for example 2.0 m.

[0063] Similarly, for other types of target objects, such as a vehicle, a pedestrian, a bicycle, a motorcycle, or the like, a height or a width or both of the sampling area may be variably adjusted or remain fixed in response to a moving speed of the target object.

[0064] Subsequently, in step S530 (means for determining the presence of a lane edge), it is determined whether or not a marking line exists in the sampling area. As shown in Fig. 7B, if a marker line (a solid line or the like, denoted by "A") is currently present in the sampling area, or if no marker line is present in the sampling area but an extension (broken line) of a marker line (a solid line or the like, denoted by "B") detected in the captured image passes through the sampling area, the collision probability correction process proceeds to step S590 (explained later) without correcting the collision probability. On the other hand, if it is determined in step S530 that a marker line is present in the sampling area, the complement of the target object is performed as shown in Fig. 1B at step S550, is read from the RAM 15, and it is determined at step S560 (means for determining a vulnerable road user) on the basis of the attribute whether the target object is a vulnerable road user or not.

[0065] For example, if the host vehicle is a passenger car, the vulnerable road user refers to a pedestrian, a two-wheeled vehicle (such as a bicycle or a motorcycle), or the like, which could be injured or damaged by the host vehicle. If, based on the addition of the target object at step S560, it is determined that the target object is a vulnerable road user, the collision probability is corrected to a much higher value (a first higher value) at step S570 (means for correcting the collision probability), and the collision probability correction process proceeds to step S590 (means for outputting the collision probability), where the corrected collision probability is output to a memory, such as the RAM 15, as a preliminary collision probability. At step S570, for example, as shown in Fig. 8, if a pedestrian is detected in the 40% range of collision probability, the collision probability is 50% (10% greater than 40%).

[0066] On the other hand, if it is determined on the basis of the complement of the target object in step S560 that the target object is not a vulnerable road user, the collision probability is corrected in step S580 (means for correcting the collision probability) to a slightly higher value (a second higher value lower than the first higher value) and the corrected collision probability is calculated as in Fig. 1B, the collision probability is stored in memory (such as RAM 15) at step S590 (means for outputting the collision probability as a preliminary collision probability). Thereafter, the collision probability correction process is terminated. At step S580, for example, the collision probability is corrected to 45% (5% greater than 40%).

[0067] That is, the collision probability is obtained by focusing on a position of the target object on the collision probability map corrected in step S530, and the collision probability read from the map is increased (or decreased in some other embodiments) by a correction value prescribed in steps S570 or S580.

[0068] After step S590, the collision probability correction process is terminated. At the same time, the collision probability map generation process is also terminated (see Fig. 3B) and the collision probability calculation process proceeds to step S220 (see Fig. 3A). At step S220, another collision probability is calculated as a function of a movement trajectory of the own vehicle and the movement speed of the target object (i.e., the relative displacement vector) in a manner similar to that disclosed in Japanese Patent No. 4026400.

[0069] Thereafter, a weighted sum or product of the collision probability calculated in step S210 and the collision probability calculated in step S220 is calculated in step S230.

[0070] The collision probability calculated in step S230 is, as in Fig. 1B as a final collision probability in a memory such as the RAM 15, whereupon the collision probability calculation process is terminated.

[0071] As in Fig. 2, the activation decision process is then carried out in step S160. Fig. Figure 9 shows a flowchart of the activation decision process performed for each of the driver assistance functions. In the present embodiment, both the collision avoidance control execution unit 24 and the alarm indicator 25 are responsible for at least one driver assistance function.

[0072] First, as in Fig. As shown in Figure 9, the collision probability calculated in the collision probability calculation process, the collision time, and the collision probability for each of the driver assist functions are acquired at step 610. Whether each driver assist function should be activated is determined based on the activation criteria for the function. The activation criteria for each driver assist function include, but are not limited to, a lower limit of the collision probability and an upper limit of the collision time for activating the driver assist function. For example, the lower limit of the collision probability for the alarm indicator 25 is set to 50%, and the lower limit of the collision probability for the collision avoidance control execution section 24 is set to 60%.

[0073] In addition, the upper limit of the collision time for the alarm indicator 25 is two seconds, and the upper limit of the collision time for a hydraulic brake pressure controlling function of the collision avoidance control executive member 24 for raising a hydraulic brake pressure is 0.6 seconds.

[0074] Each of these thresholds (the lower limit of the collision probability and the upper limit of the collision time) can be adjusted to an optimal value through experimentation and research. The microcomputer 10 operates to issue an activation command (in the form of a control signal) to each of the driver-assisting functions of the collision avoidance control executive unit 24 and the alarm indicator 25 when the collision time falls below its upper limit of the collision time, which was set in the activation criteria at step S170.

[0075] Subsequently, in step S620, an activation time point is set for each driver-assisting function, whereby the upper limit of the collision time set in the activation criteria is corrected. In particular, as shown in Fig. As shown in Figure 10, when the collision probability is less than 60%, the upper limit of the collision time set in the activation criteria is reduced by 0.5 seconds. When the collision probability increases from 60% to 100%, the correction value increases from -0.5 seconds to 0 seconds. When the collision probability reaches 100%, the correction value becomes 0 seconds.

[0076] For example, for a target object in the 80% range of collision probability, the activation time (alarm time) for triggering the alarm indicator 25 to issue an alarm is set to 1.75 seconds, which is obtained by adding -0.25 seconds (an alarm time selected from the diagram in Fig. 10 read correction value). After the correction, the alarm indicator 25 will issue the alarm at the alarm time 1.75 seconds, ie, when the collision time becomes 1.75 seconds.

[0077] It should be noted that without the correction, the alarm indicator 25 would indicate the alarm at the alarm time 2.0 seconds later. Therefore, the alarm time is delayed by 0.25 seconds.

[0078] Because the alarm activation time is delayed as the collision probability decreases, it is less likely that an unnecessary alarm will occur.

[0079] Subsequently, it is determined whether the collision probability that the host vehicle will collide with the target object is equal to or higher than the lower limit of the collision probability set in the activation criteria in step S630 (deceleration control means). If the collision probability is equal to or higher than the lower limit of the collision probability set in the activation criteria, the collision time is compared with the corrected activation time in step S640.

[0080] If it is determined in step S640 that the collision time is less than the corrected activation time, the driver assist function is switched to its ON state in step S650, and the activation decision process is terminated. If it is determined in step S640 that the collision time is equal to or greater than the corrected activation time, or if it is determined in step S630 that the collision probability that the host vehicle will collide with the target object is less than the lower limit of the collision probability set in the activation criteria, the driver assist function is switched to its OFF state in step S660, and the activation decision process is terminated. (Advantages)

[0081] In the collision avoidance system 1 described in detail above, the microcomputer 10 acquires object information including the position of an object around the host vehicle (referred to as a nearby object), and acquires vehicle state information including a traveling speed and traveling direction of the host vehicle. The microcomputer 10 generates a collision probability map (a base collision probability map or a base map in the embodiment) that shows a collision probability that the host vehicle will collide with a nearby object at each point within a range surrounding the host vehicle, based on the received vehicle state information.

[0082] The microcomputer 10 further detects a width of a traffic lane in which the host vehicle is traveling (referred to as a lane), corrects a distribution of collision probability in the collision probability map (base map) to generate a corrected collision probability map in response to the detected width of the lane, reads out a collision probability associated with the position of the nearby object from the corrected collision probability map, corrects the read out collision probability in response to the detected object information, and outputs the corrected collision probability as a collision probability that the host vehicle will collide with the nearby object.

[0083] This allows the collision probability map (i.e., the collision probability distribution) to be corrected in response to lane width, so that a collision probability of the host vehicle colliding with a nearby object located in an area where the host vehicle is more likely to move is greater than a collision probability of the host vehicle colliding with a nearby object located in an area where the host vehicle is less likely to move. Thus, an area in which the host vehicle is likely to move can be more accurately determined, leading to more accurate calculation of collision probabilities and thus to more accurate collision probabilities.

[0084] In the collision avoidance system 1 as described in detail above, the microcomputer 10 sets a width of a region in which the collision probability is constant (referred to as a constant collision probability region) so that the width of the constant collision probability region decreases with the decrease in the width of a lane.

[0085] It can be assumed that the host vehicle is more likely to move within the lane without deviating from it as the lane width decreases. Therefore, it is preferable that the width of the constant collision probability region in the collision probability map decreases, so that the host vehicle is less likely to move outside the lane. This allows the collision probability distribution in the base map to be more accurately corrected in response to the lane width.

[0086] Furthermore, in the collision avoidance system 1 described in detail above, the microcomputer 10 acquires a captured image of the surroundings of the host vehicle and detects an edge (for example, a white or yellow, solid or broken, edge-indicating line) of the host vehicle's lane from the captured image. Then, the microcomputer 10 extracts an area near and surrounding the nearby object from the captured image and determines whether or not an edge of the host vehicle's lane is within the extraction range.When an edge of the lane of the own vehicle is in the sampling range, the microcomputer 10 corrects the collision probability (read from the collision probability map corrected in response to the lane width) that the own vehicle will collide with the nearby object to a higher value and outputs the corrected collision probability.

[0087] That is, if a lane edge is located within the sampling area, the collision probability map corrected in response to the lane width is adjusted to a higher value. This allows the collision probability that the host vehicle will collide with a nearby object that is moving or located within the host vehicle's travel range to be set higher if such a nearby object can be judged to be a more dangerous nearby object (a nearby object with which the host vehicle is more likely to collide).

[0088] Furthermore, in the collision avoidance system 1 described in detail above, the microcomputer 10 detects a moving speed of the nearby object and changes dimensions (for example, at least a width and a height of a rectangular area) of an extracted area (referred to as an extraction area) from the captured image as a function of the moving speed of the nearby object.

[0089] That is, in the collision avoidance system 1 described in detail above, dimensions of the sampling area are changed as a function of the moving speed of the nearby object, so that the probability of existence of an edge of the lane in the sampling area can be changed as a function of the moving speed of the nearby object, which allows to determine the change in the collision probability in response to the moving speed of the nearby object.

[0090] In the collision avoidance system 1 described in detail above, the microcomputer 10 controls the host vehicle in response to the output collision probability. That is, in the collision avoidance system 1 described in detail above, the host vehicle is controlled based on the accurately calculated collision probability, ensuring that only necessary vehicle control is performed.

[0091] Furthermore, in the collision avoidance system 1 described in detail above, the microcomputer 10 delays the start of the vehicle control when the output collision probability is below a predetermined threshold value.

[0092] In the collision avoidance system 1 described in detail above, it is less likely that a vehicle control deemed unnecessary is performed because the start of the vehicle control is delayed when the output collision probability is below the predetermined threshold.

[0093] In the collision avoidance system 1 described in detail above, the microcomputer 10 decides whether the nearby object is a vulnerable road user or not and corrects the collision probability read from the collision probability map corrected to a higher value in response to the width of the lane to a higher value when it is determined that the nearby object is a vulnerable road user.

[0094] In the collision avoidance system 1 described in detail above, since the collision probability read from the collision probability map is corrected to a higher value when the nearby object is a vulnerable road user, the corrected collision probability is outputted in such a way that more of the driver's attention is directed to the vulnerable road user. (Other embodiments)

[0095] Some other embodiments of the present invention will now be explained, which may be designed without departing from the spirit and scope of the present invention.

[0096] In the above embodiment, the brake control is performed by the collision avoidance control executing means 24, and an alarm is issued by the alarm indicator 25 in response to the collision probability. Alternatively, to avoid a collision of the host vehicle with a nearby object in the area of ​​higher collision probability, the steering control for changing the traveling direction of the host vehicle or the seat belt retraction control may be executed.

[0097] In the above embodiment, it was assumed that the road on which the host vehicle is traveling is straight. Alternatively, the road on which the host vehicle is traveling may have curves. In the case of a curved road, the collision probability distribution can be corrected to adapt to the curvature radius of the road.

[0098] The shape of the road as well as the lane width can also be taken into account and the collision probability can be corrected so that the collision probability in the middle of the road takes the maximum value.

[0099] With this training, the probability of collision can be correctly determined even in the case of a curved road,

Claims

[1] Device (10) for calculating the collision probability for a vehicle, comprising: an object information detecting means (S110) for detecting object information including a position of an object around the own vehicle, the object being referred to as a nearby object; a vehicle state information detecting means (S120) for detecting vehicle state information including a traveling speed and a traveling direction of the own vehicle; a collision probability map generating means (S310) for generating a collision probability map based on the acquired vehicle state information, indicating a collision probability that the host vehicle will collide with a nearby object at each point of an area around the host vehicle; a traffic lane width detecting means (S410) for detecting the width of a traffic lane in which the own vehicle is moving, this traffic lane being referred to as a driving lane; collision probability distribution correcting means (S420) for correcting a collision probability distribution in the collision probability map to generate a corrected collision probability map in response to the detected lane width; a collision probability correcting means (S570, S580) for reading out a collision probability associated with the position of the nearby object from the corrected collision probability map and correcting a distribution of the collision probability in the collision probability map to generate a corrected collision probability map and correcting the read out collision probability in response to the detected object information; a collision probability output means (S590) for outputting the collision probability corrected by the collision probability correcting means (S570, S580) as a more accurate collision probability that the own vehicle will collide with the nearby object; a captured image acquiring means (S110) for acquiring a captured image of the surroundings of the own vehicle; a lane edge detection means (S320) for detecting an edge of the lane of the own vehicle in the captured image; a surrounding area extraction means (S520) for extracting an area surrounding and in the vicinity of the nearby object from the captured image and a lane edge detection means (530) for detecting the presence of a lane edge to determine whether a lane edge of the own vehicle is present in the extraction area, wherein the collision probability correction means (S570, S580) for the collision probability corrects the read collision probability to a higher value when it is determined by the lane edge detection means (S530) that an edge of the lane of the own vehicle is present in the extraction area in the vicinity of the nearby object; a movement speed detecting means (S510) for detecting a movement speed of the nearby object, wherein the movement speed of the nearby object is included in the object information of the nearby object, wherein the surrounding area extraction means (S520) is adapted to change the dimensions of an area to be extracted from the captured image (referred to as an extraction area) as a function of the moving speed of the nearby object. [2] The apparatus (10) according to claim 1, wherein the collision probability distribution correcting means (S420) sets a width of a region in which the collision probability is constant (referred to as a constant collision probability region) such that the width of the constant collision probability region decreases with a decreasing width of the lane. [3] Device (10) according to claim 1, further comprising: a means (S560) for detecting a vulnerable road user to determine whether or not the nearby object is a vulnerable road user likely to be injured by the vehicle, wherein the collision probability correction means (S570, S580) for the collision probability corrects the read collision probability to a first higher value when it is determined by the lane edge detection means (S530) that an edge of the lane of the own vehicle is present in the extraction area near the nearby object and by the vulnerable road user detection means (S560) that the nearby object is a vulnerable road user who is likely to be injured by the own vehicle, and corrects the read collision probability to a second higher value which is lower than the first higher value when it is determined by the lane edge detection means (S530) that an edge of the lane of the own vehicle is present in the extraction area near the nearby object and by the vulnerable road user detection means (S560),that the nearby object is not a vulnerable road user who is likely to be injured by the vehicle. [4] Device (10) according to one of claims 1 to 3 further comprising: Vehicle control means (S170) for controlling the own vehicle to avoid a possible collision with the nearby object in response to the collision probability outputted by the collision probability output means (S590). [5] Device (10) according to claim 4 further comprising: Delay control means (S630) for delaying the start of the collision avoidance control in response to the collision probability outputted by the collision probability output means (S590). [6] The apparatus (10) according to any one of claims 1 to 3, wherein the traffic lane width detecting means (S410) detects the width of the traffic lane on the basis of the positions of the lane edges detected by the detecting means (S320). [7] The apparatus (10) of claim 1, wherein the surrounding area extraction means (S520) changes the dimensions of an area to be extracted from the captured image as a function of the moving speed of the nearby object.

Images