JP6825836B2 - Radar accident detection device and method - Google Patents

Description

The present disclosure relates to a technique for automatically detecting an accident using radar.

There is a technology that automatically detects accidents without human intervention. Patent Document 1 discloses a technique for detecting an accident by detecting a collision sound, a brake sound, or the like.

JP-A-2010-015369

However, the technology that detects collision sounds, brake sounds, etc. and automatically detects an accident may malfunction by detecting the high-pitched voice of a child, the vibration sound of a truck bed, the barking sound of an animal, or the like.

Therefore, in order to solve the above-mentioned problems, the present disclosure aims to improve the detection accuracy by preventing malfunctions in automatically detecting an accident without human intervention.

In order to achieve the above objectives, it was decided to automatically detect the accident using radar.

Specifically, the present disclosure is a reflection that acquires information on the reflection intensity of a reflection radar signal from a road whose traffic volume is to be measured in a space whose coordinate axis is the position and speed with respect to the radar installation position. The intensity acquisition unit, the target detection unit that detects the target on the road based on the information of the reflection intensity, and the accident detection unit that detects the danger status or the accident status of the target on the road. It is a radar accident detection device characterized by being provided.

Further, in the present disclosure, regarding the reflected radar signal from the road to which the traffic volume is measured, the reflection intensity acquisition step of acquiring the reflection intensity information in the space with the position and speed as the coordinate axis based on the radar installation position. And, based on the information of the reflection intensity, a target detection step for detecting the target on the road and an accident detection step for detecting the danger situation or the accident situation of the target on the road are provided in order. It is a radar accident detection method characterized by.

According to this configuration, when an accident is automatically detected without human intervention, the detection accuracy can be improved by preventing a malfunction.

Further, in the present disclosure, the accident detection unit is based on the relative position between the mutually adjacent targets on the road or the time change of the relative position, and the danger situation of the mutually adjacent targets on the road. Alternatively, it is a radar accident detection device characterized by detecting an accident situation.

According to this configuration, it is possible to detect a dangerous situation or an accident situation related to a rear-end collision or a collision with respect to targets close to each other with high accuracy without malfunction.

Further, in the present disclosure, the accident detection unit describes the danger situation of the mutually adjacent targets on the road based on the relative speed or the time change of the relative speed between the mutually adjacent targets on the road. Alternatively, it is a radar accident detection device characterized by detecting an accident situation.

According to this configuration, it is possible to detect a dangerous situation or an accident situation related to a rear-end collision or a collision with respect to targets close to each other with high accuracy without malfunction.

Further, in the present disclosure, the accident detection unit detects the danger situation or the accident situation of each target on the road based on the position or the time change of the position of each target on the road. It is a characteristic radar accident detection device.

According to this configuration, for each target, it is possible to detect a dangerous situation or an accident situation related to traffic at a red light or stop at an intersection with high accuracy without malfunction.

Further, in the present disclosure, the accident detection unit detects a danger situation or an accident situation of each target on the road based on the speed of each target on the road or a time change of the speed. It is a characteristic radar accident detection device.

According to this configuration, it is possible to detect a dangerous situation or an accident situation related to overspeed, insufficient speed, sudden deceleration or sudden acceleration, etc. for each target with high accuracy without malfunction.

Further, the present disclosure further includes a video recording command unit that issues a command to the video recording device to record the dangerous situation or the accident situation of the target on the road as a video. It is an accident detection device.

According to this configuration, it is possible to record and verify a dangerous situation or an accident situation as a video.

Further, in the present disclosure, the target detection unit uses the reflection intensity information as a reflection derived from a target whose traffic volume is to be measured in a space whose coordinate axes are the position and speed with respect to the radar installation position. The position and speed of the signal are detected as a target signal, and the target signal is formed into a cluster according to the proximity of at least one of the position and speed of the target signal to form a target on the road. Is a radar accident detection device characterized by detecting as the cluster.

According to this configuration, targets such as large cars, small cars, pedestrians, and bicycles are not measured as points without size, but as clusters with size. Therefore, it is possible to discriminate targets such as large vehicles, small vehicles, pedestrians, and bicycles based on the size and speed of the cluster. Then, when the accident is automatically detected without human intervention, the detection accuracy can be further improved by further preventing the malfunction.

As described above, according to the present disclosure, it is possible to improve the detection accuracy by preventing a malfunction in automatically detecting an accident without manpower.

It is a block diagram which shows the structure of the radar accident detection apparatus of this disclosure. It is a flowchart which shows the process of the radar accident detection apparatus of this disclosure. It is a figure which shows the process of the target signal detection of this disclosure. It is a figure which shows the process of the target signal detection of this disclosure. It is a figure which shows the clustering process of this disclosure. It is a figure which shows the clustering process of this disclosure. It is a figure which shows the clustering process of this disclosure. It is a figure which shows the process of the cluster locus extraction of this disclosure. It is a figure which shows the process of the relative position detection of this disclosure. It is a figure which shows the process of the relative velocity detection of this disclosure. It is a figure which shows the process of the position detection of this disclosure. It is a figure which shows the process of speed detection of this disclosure.

Embodiments of the present disclosure will be described with reference to the accompanying drawings. The embodiments described below are examples of the embodiments of the present disclosure, and the present disclosure is not limited to the following embodiments.

The configuration and processing of the radar accident detection device of the present disclosure are shown in FIGS. 1 and 2. Radar sensors (not shown) are installed in road equipment such as traffic lights, display boards, and telephone poles. The radar accident detection device D receives the reflected radar signal from the radar sensor, and transmits a command to record the danger situation or the accident situation as an image to the image recording device (not shown) via a network (not shown). The dangerous situation refers to a situation in which an accident has not actually occurred, but an accident may occur. In addition, the accident situation refers to the situation in which the accident actually occurs.

The radar accident detection device D is composed of a reflection intensity acquisition unit 1, a target detection unit 2, an accident detection unit 3, and a video recording command unit 4, and steps S1 to S4 are executed in this order.

The reflection intensity acquisition unit 1 acquires information on the reflection intensity of the reflection radar signal from the road whose traffic volume is to be measured in a space whose coordinate axis is the position and speed with respect to the radar installation position (step S1). ). The distance of the radar reflection point from the radar sensor is measured based on the propagation delay of the reflection radar signal, and the orientation of the radar reflection point with respect to the radar sensor is based on the phase change of the reflection radar signal between multiple antennas. Measured and based on the phase change of the reflected radar signal between multiple sweeps, the speed of the radar's reflection point with respect to the radar sensor is measured.

The target detection unit 2 detects a target on the road based on the above reflection intensity information (step S2). Here, the target detection unit 2 is composed of a target signal detection unit 21, a clustering unit 22, and a cluster locus extraction unit 23, and steps S21 to S23 are executed in this order.

The target signal detection unit 21 of the target detection unit 2 derives the above reflection intensity information from the target for measuring the traffic volume in the space with the position and speed as the reference axis of the radar installation position. The position and speed of the reflected signal are detected as a target signal (step S21). The processing of the target signal detection of the present disclosure is shown in FIGS. 3 and 4.

The target signal detection unit 21 dynamically sets the detection thresholds T1 and T2 of the target signal according to the reflection intensity of the background such as road surface reflection, and detects the intensity peaks P1 and P2. Here, the higher the assumed ratio of the traffic volume of the large vehicle to the traffic volume of the small vehicle, the lower the detection threshold T2 of the target signal, and the lower the assumed ratio of the traffic volume of the large vehicle to the traffic volume of the small vehicle, the higher the target. The signal detection threshold T1 is raised to detect the above target signal.

That is, for a small car, the radar signal is less likely to be specularly reflected on the surface of the vehicle body, so that even if the detection threshold value T1 of the target signal is increased, the positions of the target signal derived from the small car are rarely separated. On the other hand, for large vehicles, the radar signal is often specularly reflected on the ceiling of the loading platform. Therefore, if the detection threshold value T1 of the target signal is increased, the positions of the target signals derived from the large vehicle are often separated. Therefore, as shown in FIG. 3, by lowering the detection threshold value T2 of the target signal, it is possible to prevent a large vehicle from being erroneously determined as a small vehicle.

The clustering unit 22 of the target detection unit 2 forms a cluster for the above target signal according to the proximity of at least one of the position and speed of the target signal (step S22). The clustering process of the present disclosure is shown in FIGS. 5 to 7. The target signals in FIGS. 5 to 7 are shown as a small number of point clouds for the sake of simplicity in the drawings.

Specifically, as shown in FIG. 5, the clustering unit 22 collects the target signals of seven points whose positions and speeds are close to each other as cluster C. Then, the cluster C, the size L _X of the traveling direction and the vertical _direction, the size L _Y direction of travel and parallel to the _direction, the center position R _C and characteristic velocity V _{C (e.g.,} the average speed of a target signal of 7 points Value) is measured.

Here, in the clustering unit 22, the lower the distribution density of the target signal in the space with the position based on the radar installation position as the coordinate axis, and / or the faster the speed of the target signal, the more the target signal of the target signal. The cluster may be formed by placing more weight on the proximity of the target signal velocity than on the proximity of the position.

Specifically, as shown in FIG. 6, the clustering unit 22 collects the target signals of the upper three points and the target signals of the middle three points, whose positions are not close to each other and their velocities are close to each other, as cluster C1. .. On the other hand, the target signals at the lower three points whose positions are not close to each other and their velocities are not close to those at the three points in the middle stage are grouped as cluster C2. Then, the cluster C1, for C2, respectively, the traveling direction and the vertical size _{L X1, L X2,} the size of the running direction and parallel to the direction _{L Y1, L Y2,} center position _{R C1, R C2} and characteristic velocity _{V C1} , _VC2 is measured.

Here, when the distribution density of the target signal is low and / or the speed of the target signal is high in non-congestion, the speed varies from vehicle to vehicle. Therefore, as shown in FIG. 6, a large vehicle is erroneously determined as a small vehicle by performing clustering by placing weight on the proximity of the target signal speed rather than the proximity of the target signal position. Can be prevented.

On the other hand, in the clustering unit 22, the higher the distribution density of the target signal in the space with the position based on the radar installation position as the coordinate axis, and / or the slower the speed of the target signal, the more the target signal of the target signal. The cluster may be formed by placing more weight on the proximity of the target signal position than on the proximity of the velocity.

Specifically, as shown in FIG. 7, the clustering unit 22 outputs the target signals of the upper three points and the target signals of the middle three points, which are close in velocity but not close in position, to the cluster C3, respectively. , C4. On the other hand, the target signals at the lower three points, which are close in velocity but not close to the target signals at the three points in the middle stage, are grouped as cluster C5. Then, the cluster C3, the C4, C5, respectively, the magnitude of the traveling direction and the vertical direction _{L X3, L X4, L X5} , the magnitude of the traveling direction and parallel to the direction _{L Y3, L Y4, L Y5} , center position _{R C3} , _RC4 , _RC5 and representative speeds _VC3 , _VC4 , _VC5 are measured.

Here, when the distribution density of the target signal is high and / or the speed of the target signal is slow in a traffic jam, the speed of each vehicle has little variation. Therefore, as shown in FIG. 7, by performing clustering by placing weight on the degree of proximity of the target signal position rather than the degree of proximity of the target signal speed, one adjacent vehicle having a similar speed is used. It is possible to prevent erroneous determination as a vehicle of.

The cluster locus extraction unit 23 of the target detection unit 2 extracts the locus of the above cluster in the space with the position based on the radar installation position as the coordinate axis, and the speed of the above cluster having no movement in the locus. It is considered to be 0, and the above-mentioned cluster having motion in its trajectory is regarded as having a non-zero velocity (step S23). The process of extracting the cluster locus of the present disclosure is shown in FIG.

The cluster locus extraction unit 23 determines that there is a time correlation with respect to the position, velocity (finite value), and magnitude in the clusters C6 to C8 at times t1 to t3. Then, the clusters C6 to C8 at times t1 to t3 are summarized as a locus of one target. Furthermore, since there is movement in the trajectory of that one target, it is considered that the velocity is not zero.

The cluster locus extraction unit 23 determines that there is a time correlation with respect to the position, velocity (almost 0) and magnitude in the clusters C9 to C11 at times t1 to t3. Then, the clusters C9 to C11 at times t1 to t3 are summarized as a locus of one target. Further, since there is no movement in the locus of the one target, the velocity is considered to be zero.

The accident detection unit 3 detects a dangerous situation or an accident situation of a target on the road (step S3). Here, the accident detection unit 3 is composed of a relative position detection unit 31, a relative speed detection unit 32, a position detection unit 33, and a speed detection unit 34, and steps S31 to S34 are executed in no particular order.

The relative position detection unit 31 of the accident detection unit 3 determines the danger situation or accident of the mutually adjacent targets on the road based on the time change of the relative position or the relative position between the mutually adjacent targets on the road. Detect the situation (step S31). The process of relative position detection of the present disclosure is shown in FIG.

First, a dangerous situation or an accident situation in which targets close to each other collide with each other will be described. Relative position detecting unit 31, the relative distance between the target object corresponding to the cluster C12, C13 is when it becomes Y ₁ is less than a predetermined value, hazardous situation or accident situation regarding collision is judged to have occurred. Alternatively, when the relative distance between the target object corresponding to the cluster C12, C13 is narrowed from Y ₁ by a predetermined value or more rate of change to Y _2, it is determined that the hazardous situation or accident situation regarding collision occurs.

Next, a dangerous situation or an accident situation in which objects close to each other collide with each other will be described. Relative position detecting unit 31, the relative distance between the target object corresponding to the cluster C14, C15 is when it becomes Y ₃ is less than the predetermined value, it is determined that the hazardous situation or accident situation regarding collision occurs. Alternatively, when the relative distance between the target object corresponding to the cluster C14, C15 is narrowed from Y ₃ at a predetermined value or more rate of change to Y _4, it is determined that the hazardous situation or accident situation regarding collision occurs.

Here, in the detection process of the danger situation or the accident situation by the relative position detection unit 31, clustering is performed by placing more weight on the proximity of the target signal position than on the proximity of the target signal speed. When it is, it is more effective. Therefore, as described with reference to FIG. 7, it is more effective in a traffic jam when there is not enough distance between targets that are close to each other.

The relative speed detection unit 32 of the accident detection unit 3 determines the danger situation or accident of the mutually adjacent targets on the road based on the relative speed or the time change of the relative speed between the mutually adjacent targets on the road. Detect the situation (step S32). The process of relative velocity detection of the present disclosure is shown in FIG.

First, a dangerous situation or an accident situation in which targets close to each other collide with each other will be described. The relative velocity detection unit 32 becomes V ₁ in which the relative velocity between the targets corresponding to the clusters C16 and C17 (however, the relative velocity becomes positive when the relative distance is narrowed) is equal to or more than a predetermined value. At that time, it is judged that a dangerous situation or an accident situation related to the collision has occurred. Alternatively, the relative velocity between the target object corresponding to the cluster C16, C17 is when increased from V ₁ to V ₂ at a predetermined value or more rate of change, it is determined that the hazardous situation or accident situation regarding collision occurs.

Next, a dangerous situation or an accident situation in which objects close to each other collide with each other will be described. Relative speed detection unit 32, the relative velocity between the target object corresponding to the cluster C18, C19 (provided that when the relative distance is narrowed, the relative velocity is positive.) Becomes V ₃ is equal to or higher than a predetermined value At that time, it is judged that a dangerous situation or an accident situation related to the collision has occurred. Alternatively, the relative velocity between the target object corresponding to the cluster C18, C19 is when increased from V ₃ to V ₄ at a predetermined value or more rate of change, it is determined that the hazardous situation or accident situation regarding collision occurs.

Here, in the detection process of the danger situation or the accident situation by the relative speed detection unit 32, clustering is performed by placing more weight on the degree of proximity of the target signal speed than on the degree of proximity of the target signal position. When it is, it is more effective. Therefore, as described with reference to FIG. 6, it is more effective in the non-congested state where there is a margin between the targets that are close to each other.

The position detection unit 33 of the accident detection unit 3 detects the danger situation or the accident situation of each target on the road based on the position or the time change of the position of each target on the road (step S33). The position detection process of the present disclosure is shown in FIG.

First, a dangerous situation or an accident situation in which each target passes at a red light will be explained. The position detection unit 33 determines that a dangerous situation or an accident situation related to traffic at the red light has occurred when the position of the target corresponding to the cluster C20 is on the road where the red light appears. Alternatively, it is determined that a dangerous situation or an accident situation related to traffic at the red light has occurred when the position of the target corresponding to the cluster C20 travels on the road where the red light appears.

Next, a dangerous situation or an accident situation in which each target stops at an intersection will be described. The position detection unit 33 determines that a dangerous situation or an accident situation related to a stop in the intersection has occurred when the position of the target corresponding to the cluster C21 exists in the intersection for a predetermined time or more.

Here, in the detection process of the danger situation or the accident situation by the position detection unit 33, clustering is performed by placing more weight on the degree of proximity of the target signal position than on the degree of proximity of the target signal speed. Sometimes it's more effective.

The speed detection unit 34 of the accident detection unit 3 detects the danger situation or the accident situation of each target on the road based on the speed of each target on the road or the time change of the speed (step S34). The speed detection process of the present disclosure is shown in FIG.

First, a dangerous situation or an accident situation in which the speed of each target is exceeded will be described. Speed detecting unit 34, when the speed of the target object corresponding to the cluster C22 becomes V ₅ is not less than a predetermined value, hazardous situation or accident situation regarding Overspeed is judged to have occurred.

Next, a dangerous situation or an accident situation in which the speed of each target is insufficient will be described. Speed detecting unit 34, when the speed of the target object corresponding to the cluster C23 becomes V ₆ is less than the predetermined value, hazardous situation or accident situation regarding insufficient speed is judged to have occurred.

Next, a dangerous situation or an accident situation in which each target suddenly decelerates will be described. Speed detecting unit 34, when the speed of the target object corresponding to the cluster C24 is reduced to V ₈ from V ₇ at a predetermined value or more rate of change, it is determined that the hazardous situation or accident situation regarding the sudden deceleration occurs.

Next, a dangerous situation or an accident situation in which each target accelerates rapidly will be described. The speed detection unit 34 determines that a dangerous situation or an accident situation related to sudden acceleration has occurred when the speed of the target corresponding to the cluster C25 increases from V ₉ to V ₁₀ at a rate of change equal to or higher than a predetermined value.

Here, in the detection process of the danger situation or the accident situation by the speed detection unit 34, clustering is performed by placing more weight on the proximity of the speed of the target signal than on the proximity of the position of the target signal. Sometimes it's more effective.

The video recording command unit 4 issues a command to record a dangerous situation or an accident situation of a target on the road as a video to a video recording device (not shown) (step S4). As a result of the above processing, the dangerous situation or the accident situation can be recorded and verified as a video.

The radar accident detection device and method of the present disclosure can improve the detection accuracy by preventing a malfunction when automatically detecting an accident without human intervention.

D: Radar accident detection device 1: Reflection intensity acquisition unit 2: Target detection unit 3: Accident detection unit 4: Video recording command unit 21: Target signal detection unit 22: Clustering unit 23: Cluster trajectory extraction unit 31: Relative position Detection unit 32: Relative velocity detection unit 33: Position detection unit 34: Speed detection unit P1, P2: Intensity peak T1, T2: Detection threshold C, C1 to C25: Cluster

Claims (7)

Regarding the reflected radar signal from the road to which the traffic volume is measured, the reflection intensity acquisition unit that acquires the reflection intensity information in the space with the position and speed as the reference axis of the radar installation position, and
A target detection unit that detects a target on the road based on the reflection intensity information,
An accident detection unit that detects the danger situation or accident situation of the target on the road,
Equipped with a,
The target detection unit
Regarding the information on the reflection intensity, the position and speed of the reflected signal derived from the target for which the traffic volume is to be measured are detected as the target signal in the space whose coordinate axis is the position and speed based on the radar installation position. With respect to the target signal, a cluster is formed according to the proximity of at least one of the position and speed of the target signal, and the target on the road is detected as the cluster.
The lower the distribution density of the target signal and / or the faster the speed of the target signal, the closer the weight is to the speed of the target signal rather than the proximity of the position of the target signal. The cluster is formed, and the higher the distribution density of the target signal and / or the slower the speed of the target signal, the closer the speed of the target signal is to the closer the target signal is. A radar accident detection device characterized in that the cluster is formed by placing weight on the proximity of positions . The accident detection unit detects the danger situation or the accident situation of the mutually adjacent targets on the road based on the relative position or the time change of the relative position between the mutually adjacent targets on the road. The radar accident detection device according to claim 1, wherein the radar accident detection device is characterized in that. The accident detection unit detects the danger situation or the accident situation of the mutually adjacent targets on the road based on the relative speed or the time change of the relative velocity between the mutually adjacent targets on the road. The radar accident detection device according to claim 1 or 2, wherein the radar accident detection device is characterized in that. The claim is characterized in that the accident detection unit detects a danger situation or an accident situation of each target on the road based on a position or a time change of the position of each target on the road. The radar accident detection device according to any one of 1 to 3. The claim is characterized in that the accident detection unit detects a danger situation or an accident situation of each target on the road based on the speed of each target on the road or a time change of the speed. The radar accident detection device according to any one of 1 to 4. Claims 1 to 5, further comprising a video recording command unit that issues a command to the video recording device to record a dangerous situation or an accident situation of a target on the road as a video. The radar accident detection device described in either. Regarding the reflected radar signal from the road to be measured for traffic volume, the reflection intensity acquisition step for acquiring the reflection intensity information in the space with the position and speed as the reference axis of the radar installation position, and
A target detection step for detecting a target on the road based on the reflection intensity information,
An accident detection step for detecting a dangerous situation or an accident situation of a target on the road,
In order ,
The target detection step
Regarding the information on the reflection intensity, the position and speed of the reflected signal derived from the target for which the traffic volume is to be measured are detected as the target signal in the space whose coordinate axis is the position and speed based on the radar installation position. With respect to the target signal, a cluster is formed according to the proximity of at least one of the position and speed of the target signal, and the target on the road is detected as the cluster.
The lower the distribution density of the target signal and / or the faster the speed of the target signal, the closer the weight is to the speed of the target signal rather than the proximity of the position of the target signal. The cluster is formed, and the higher the distribution density of the target signal and / or the slower the speed of the target signal, the closer the speed of the target signal is to the closer the target signal is. A radar accident detection method characterized in that the cluster is formed by placing weight on the proximity of positions .

Images