JP2019207655A - Detection device and detection system - Google Patents

Abstract

To provide a detection device which offers improved red light violation detection accuracy.SOLUTION: A detection device provided herein is configured to detect red light violation using a laser radar, and comprises a first acquisition unit for acquiring point cloud information, an object detection unit configured to detect an object using the point cloud information, an object tracking unit configured to track the object detected by the object detection unit and generate a tracking result, a second acquisition unit for acquiring traffic light information representing the state of a traffic light, and a behavior detection unit configured to detect red light violation committed by the object on the basis of the tracking result and the traffic light information. The behavior detection unit determines that the object has committed red light violation when it determines that the object entered an intersection while the traffic light was signaling vehicles to stop.SELECTED DRAWING: Figure 1

Description

  The present disclosure relates to a detection device and a detection system.

  Conventionally, a technology for detecting a signal-ignoring vehicle is known. For example, in Patent Document 1, a signal state detection unit detects a signal state, a laser distance meter detects a vehicle that has entered a crossing over a main road stop line, and a vehicle that has entered the crossing at the time of a red signal Is detected as a signal ignoring vehicle.

JP 2003-141679 A

  The apparatus described in Patent Document 1 detects that a vehicle has entered an intersection when the laser beam irradiated by the laser rangefinder is blocked by the vehicle. However, when the main road has a red light, the laser beam may be blocked by a bicycle or a pedestrian crossing a pedestrian crossing intersecting the main road. In this case, ignoring the signal of the vehicle may be erroneously detected.

  The present disclosure describes a detection device capable of improving detection accuracy of signal ignorance.

  A detection device according to one aspect of the present disclosure uses a laser radar that generates point cloud information by irradiating a detection area set at an intersection provided with a traffic light with a laser beam and receiving reflected light of the laser beam. This is a device for detecting signal ignorance. The detection device includes a first acquisition unit that acquires point cloud information, an object detection unit that detects an object based on the point cloud information, an object that tracks the object detected by the object detection unit, and generates a tracking result A tracking unit; a second acquisition unit that acquires signal information indicating a signal state of a traffic light that gives an instruction regarding the progress of the object; and a behavior detection unit that detects signal ignorance of the object based on the tracking result and the signal information. Prepare. When the behavior detection unit determines whether the object has entered the intersection when the traffic light is in a state indicating stop, and determines that the object has entered the intersection when the traffic light is in a state indicating stop, It is determined that the object has ignored the signal.

  According to the present disclosure, it is possible to improve the detection accuracy of signal ignorance.

FIG. 1 is a diagram schematically illustrating a configuration of a detection system including a detection device according to an embodiment. FIG. 2 is a diagram showing an arrangement example of the laser radar and the vehicle information acquisition device shown in FIG. FIG. 3 is a diagram illustrating a hardware configuration of the detection device illustrated in FIG. 1. FIG. 4 is a diagram for explaining signal ignorance. FIG. 5 is a flowchart showing a series of processes of a detection method performed by the detection apparatus. FIG. 6 is a flowchart showing in detail the vehicle behavior detection process of FIG. FIG. 7 is a flowchart showing in detail the analysis processing of FIG. FIG. 8 is a flowchart showing in detail the deletion process of FIG. FIG. 9 is a diagram illustrating an example of the management table. FIG. 10 is a diagram for explaining a modified example of signal ignore detection. FIG. 11 is a flowchart showing in detail the analysis processing in the modification shown in FIG. FIG. 12 is a diagram showing an example of the management table in the modification shown in FIG. FIG. 13 is a diagram for explaining the passage of the detection line. FIG. 14 is a diagram for explaining signal ignorance in the modification shown in FIG. FIG. 15 is a diagram for explaining the progress in a specific direction in the modification shown in FIG.

[1] Outline of Embodiments A detection device according to one aspect of the present disclosure irradiates a detection area set at an intersection provided with a traffic light with a laser beam and receives reflected light of the laser beam to obtain point cloud information. This is a device that detects signal ignorance using a laser radar that generates The detection device includes a first acquisition unit that acquires point cloud information, an object detection unit that detects an object based on the point cloud information, an object that tracks the object detected by the object detection unit, and generates a tracking result A tracking unit; a second acquisition unit that acquires signal information indicating a signal state of a traffic light that gives an instruction regarding the progress of the object; Prepare. When the behavior detection unit determines whether the object has entered the intersection when the traffic light is in a state indicating stop, and determines that the object has entered the intersection when the traffic signal is in a state indicating stop, It is determined that the object has ignored the signal.

  In this detection apparatus, an object is detected using the point cloud information generated by the laser radar, and the detected object is tracked. Based on the tracking result of the object and the signal information indicating the state of the traffic light, it is determined whether or not the object has entered the intersection when the traffic light is in the stop state, and the traffic light is in the stop state. When it is determined that the object has entered the intersection, it is determined that the object has ignored the signal. By tracking the object, the movement trajectory of the object is obtained. A pedestrian or the like crossing a road through a pedestrian crossing moves in a different direction from a vehicle entering the intersection from the road. For this reason, by tracking an object, it is possible to reduce the possibility of erroneous detection of a pedestrian crossing a pedestrian crossing as a vehicle entering the intersection. As a result, it is possible to improve the detection accuracy of ignoring signals.

  When the signal information further indicates that the traffic signal indicates a direction in which the traffic signal can pass, the behavior detection unit compares the traveling direction of the object with the direction in which the traffic can pass, so that the object ignores the signal. It may be determined whether or not. There is a traffic light indicating that a specific direction is possible while indicating a stop. At the intersection where such a traffic light is installed, it is not determined that the object has ignored the signal when the object travels in the passable direction by comparing the travel direction of the object with the passable direction. Can be.

  Detection lines for detecting that an object has left the intersection may be set at boundaries between the intersection and a plurality of roads extending from the intersection. The behavior detection unit may specify the traveling direction when the object exceeds the detection line. When an object leaves the intersection, the object always exceeds the detection line set at the boundary between the intersection and one of the roads. For this reason, by using the detection line, it is possible to reliably detect the exit from the intersection regardless of the speed of the object, and it is possible to improve the accuracy of specifying the traveling direction. As a result, it is possible to further improve the detection accuracy of ignoring signals.

  The behavior detection unit may classify the object according to the signal state indicated by the signal information when the object enters the intersection. The danger varies depending on the signal state when the object enters the intersection. For this reason, a warning etc. can be given according to danger by classifying an object according to a signal state.

  The behavior detection unit may output an acquisition command for causing an external device to acquire object information for specifying an object for which the signal is ignored. In this case, object information for specifying the object for which the signal is ignored is acquired. For this reason, for example, it is possible to give violation points and collect fines for the driver of the vehicle who has ignored the signal.

  A detection system according to another aspect of the present disclosure includes the detection device described above and a laser radar. Since this detection system includes the above-described detection device, it is possible to improve the detection accuracy of signal ignorance.

[2] Examples of Embodiments Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a diagram schematically illustrating a configuration of a detection system including a detection device according to an embodiment. FIG. 2 is a diagram showing an arrangement example of the laser radar and the vehicle information acquisition device shown in FIG. FIG. 3 is a diagram illustrating a hardware configuration of the detection device illustrated in FIG. 1. FIG. 4 is a diagram for explaining signal ignorance.

  A detection system 1 shown in FIG. 1 is a system that detects ignorance of a vehicle signal at a target point. The target point is a place (point) on the road to be monitored and is a place where the traffic light TL is provided. An example of the target point is an intersection. Examples of vehicles include automobiles and motorcycles. The detection system 1 is used for an intelligent transport system (ITS). The detection system 1 includes a laser radar 2, a vehicle information acquisition device 3 (external device), an output device 4, a storage device 5, a communication device 6, and a detection device 10.

  The laser radar 2 is a device that generates point cloud information. The laser radar 2 is also called a lidar (LiDAR: Light Detection And Ranging) or a Laser Range Finder. As shown in FIG. 2, the laser radar 2 is installed in the vicinity of the target point. In the example shown in FIG. 2, the target point is an intersection P. At the intersection P, a traffic light TL is provided that gives an instruction regarding the traveling of the vehicle to the vehicle traveling on the road TR. A stop line Ls is provided in front of the intersection P. The laser radar 2 is provided above the target point, for example. The laser radar 2 is fixed to a support member 7 installed on the ground. The support member 7 is provided on the road side of the road TR, for example. The support member 7 is, for example, a support column erected on the road side. The support member 7 may be, for example, a utility pole and a warehouse wall.

  The laser radar 2 irradiates laser light toward the irradiable range Ra and generates point cloud information by receiving reflected light of the irradiated laser light. The irradiable range Ra is a range in which the laser radar 2 can irradiate laser light, for example, a range of about 150 m. The irradiation possible range Ra includes a detection area Rd. The detection area Rd is an area to be monitored in the irradiable range Ra. In the example shown in FIG. 2, the detection area Rd is set at the intersection P. The point cloud information is a set of measurement point information of each measurement point included in the irradiable range Ra.

  The measurement point information includes time information and position information. The time information is information indicating the time when the measurement point information is generated (received reflected light) for the measurement point indicated by the position information. The position information is information indicating the position of the measurement point. For the position, a polar coordinate system represented by a yaw angle, a pitch angle, and a depth may be used, or a three-dimensional coordinate system of an X coordinate, a Y coordinate, and a Z coordinate may be used. As the coordinates of each measurement point, for example, coordinates with the installation position of the laser radar 2 as the origin are used. The measurement point information may further include reflection intensity information. The reflection intensity information is information indicating the intensity of reflected light received at the time indicated by the time information from the measurement point indicated by the position information.

  The laser radar 2 scans the irradiable range Ra in the main scanning direction and the sub-scanning direction using laser light. Point cloud information obtained by irradiating the irradiable range Ra with laser light may be referred to as one frame. Irradiation of the laser beam to the irradiable range Ra is repeated at predetermined time intervals. The laser radar 2 transmits the generated point cloud information to the detection device 10.

  The vehicle information acquisition device 3 is a device that acquires vehicle information for specifying a vehicle that has been determined to have been ignored by the detection device 10. The vehicle information acquisition device 3 is, for example, an automatic speed violation control device. As the vehicle information, for example, an image of a license plate is used. When the vehicle information acquisition device 3 receives the acquisition command together with the vehicle position information from the detection device 10, for example, the vehicle information acquisition device 3 performs imaging so as to include a license plate of a vehicle (acquisition target vehicle) that exists at the position indicated by the vehicle position information. Get license plate images.

  When the vehicle to be acquired enters the monitoring range Rb, the vehicle information acquisition device 3 captures the vehicle. The monitoring range Rb is a range in which the vehicle information acquisition device 3 can shoot, and is about several meters. In the example shown in FIG. 2, the monitoring range Rb is set so that the vehicle information acquisition device 3 can photograph the vehicle from the front of the vehicle.

  Similarly to the laser radar 2, the vehicle information acquisition device 3 is fixed to a support member. The laser radar 2 and the vehicle information acquisition device 3 may be attached to the same support member, or may be attached to different support members. In FIG. 2, the support member is not shown. The vehicle information acquisition device 3 may transmit the vehicle information to the storage device 5 and store the vehicle information in the storage device 5. The vehicle information acquisition device 3 may transmit vehicle information to a public organization, a road management company, or the like via the communication device 6.

  The output device 4 is a device that performs caution, warning, and instruction. The output device 4 includes, for example, a display board and a speaker. When the output device 4 receives the output command from the detection device 10, the output device 4 outputs cautions, warnings, instructions, and the like to the driver of the violating vehicle. The output device 4 may output cautions, warnings, instructions, and the like toward surrounding pedestrians and the like.

  The storage device 5 is a device that stores various types of information. Examples of the storage device 5 include a hard disk device and a semiconductor memory. As various information memorize | stored in the memory | storage device 5, the vehicle information acquired by the vehicle information acquisition apparatus 3, the detection time of the detection vehicle mentioned later, a position, speed, and a magnitude | size (dimension) are mentioned, for example. When the vehicle information acquisition device 3 includes a video camera, the various types of information may include a video movie in a time zone including the detection time. The various types of information may include point cloud information in a time zone including the detection time. The various types of information may include the movement trajectory (time-series vehicle position information) of the detected vehicle.

  The communication device 6 is a communication facility installed on the side of the road. The communication device 6 is, for example, a roadside machine (ITS spot). The communication device 6 communicates with a device outside the detection system 1. Examples of the external device include an in-vehicle device, a server device, and other systems. The communication device 6 transmits information similar to various information stored in the storage device 5 to an external device.

  The detection device 10 is a device for detecting neglect of a vehicle signal in the detection area Rd using the laser radar 2. The detection device 10 is configured by an information processing device such as a computer, for example.

  As shown in FIG. 3, the detection device 10 physically includes one or more processors 101, a main storage device 102 such as a RAM (Random Access Memory) and a ROM (Read Only Memory), and an auxiliary device such as a hard disk device. The storage device 103, an input device 104 such as a keyboard, an output device 105 such as a display, and a computer including hardware such as a communication device 106 that is a data transmission / reception device may be configured. Each function shown in FIG. 1 of the detection device 10 is performed by causing each hardware component such as the main storage device 102 to read one or more predetermined computer programs under the control of one or more processors 101. This is realized by operating the hardware and reading and writing data in the main storage device 102 and the auxiliary storage device 103.

  Functionally, the detection device 10 includes an acquisition unit 11 (first acquisition unit), a setting unit 12, an object detection unit 13, an object tracking unit 14, a behavior detection unit 15, and an acquisition unit 16 (second Acquisition unit).

  The acquisition unit 11 acquires point cloud information from the laser radar 2. The acquisition unit 11 outputs the acquired point cloud information to the object detection unit 13.

  The setting unit 12 acquires area information indicating the detection area Rd. For example, the user sets the detection area Rd using the input device 104 of the detection device 10. For example, a three-dimensional space simulating the target point is displayed on the output device 105, and the user sets the detection area Rd with a frame or the like. For example, the detection area Rd is set so as not to include a sidewalk. In addition, there are fixed objects such as traffic lights, poles, utility poles, and overheads at the intersection. Since these fixed objects are not monitored, in order to exclude these fixed objects, a range of a predetermined height or less may be set as the height of the detection area Rd. In addition, in order to exclude the ground, a range of a predetermined height or more may be set as the height of the detection area Rd. For example, the height of the detection area Rd may be set in a range from 20 cm to 500 cm with reference to the ground surface. The setting unit 12 acquires area information indicating the set detection area Rd, and outputs the acquired area information to the object detection unit 13.

  The object detection unit 13 detects an object based on the point cloud information. Specifically, when the object detection unit 13 receives the point cloud information from the acquisition unit 11, the object detection unit 13 obtains the measurement point information of the measurement points located outside the detection area Rd indicated by the area information received from the setting unit 12. Exclude from The object detection unit 13 clusters the remaining point group information. That is, the object detection unit 13 connects nearby measurement points among a plurality of measurement points in the detection area Rd, and divides them into clusters (lumps). The object detection unit 13 detects the obtained cluster as a single detection object (such as a vehicle and a person).

  The object detection unit 13 calculates the size (width, depth, and height) and position of the detected object. The position of the detection object may be the coordinates of the four corners (front right end, front left end, rear right end, and rear left end) of the detection object, may be the average of the positions of the measurement point information included in the cluster, or may be the barycentric position of the detection object . The position of the detection object may be the coordinates of the front right end and the front left end, or may be the coordinates of the measurement point closest to the laser radar 2.

  Since the detection device 10 detects the behavior of the vehicle, it is not necessary to track a moving object (non-vehicle) such as a person or a bird. For this reason, the object detection part 13 excludes a non-vehicle from a detection object. Specifically, the object detection unit 13 classifies the detected object based on the size (dimension) of the detected object. Here, the object detection unit 13 classifies the detected object into a vehicle and a non-vehicle. For example, the object detection unit 13 classifies the detected object as a non-vehicle when the width of the detected object is less than 1 m, and classifies the detected object as a vehicle when the width of the detected object is 1 m or more.

  The object detection unit 13 outputs the detection result to the object tracking unit 14 for the detected object classified as a vehicle (hereinafter referred to as “detected vehicle”). The detection result includes size information indicating the size of the detected vehicle, vehicle position information indicating the position of the detected vehicle, and detection time information indicating the detection time when the detected vehicle is detected. The detection time is, for example, an average time indicated by time information included in measurement point information of each measurement point included in the cluster.

  The object tracking unit 14 tracks the detected vehicle and generates a tracking result. That is, when the object tracking unit 14 receives the detection result from the object detection unit 13, the object tracking unit 14 associates a vehicle ID (Identifier) with the detected vehicle detected in a different frame (different detection time). The vehicle ID is identification information that can uniquely identify the detected vehicle. Specifically, the object tracking unit 14 detects the detected vehicle detected in the current frame based on the position and size of the detected vehicle and the speed and angular velocity estimated from the past observation results. It is determined whether it corresponds to one of the detected vehicles detected in step (1).

  When the object tracking unit 14 determines that the detected vehicle detected in the current frame does not correspond to any of the detected vehicles detected in the past frame, the object tracking unit 14 assigns a new vehicle ID to the detected vehicle as a new detected vehicle. Give. When the object tracking unit 14 determines that the detected vehicle detected in the current frame corresponds to the detected vehicle detected in the past frame, the object tracking unit 14 sets the vehicle ID assigned to the corresponding detected vehicle in the current frame. It is given to the detected vehicle. The object tracking part 14 deletes vehicle ID about the detection vehicle which has not been detected for a long time among the detection vehicles to which vehicle ID is provided.

  The problem of tracking (identifying) a plurality of detected vehicles is referred to as a multi-target tracking problem. The object tracking unit 14 tracks each detected vehicle using a known algorithm. Known algorithms include SNN (Suboptimal Nearest Neighbor), GNN (Global Nearest Neighbor), JPDAF (Joint Probabilistic Data Association Filter), and the like. The object tracking unit 14 outputs the tracking result to the behavior detecting unit 15. The tracking result includes a vehicle ID, vehicle position information, and detection time information.

  The acquisition unit 16 acquires signal information. The signal information is information indicating the signal state of the traffic light TL that gives an instruction regarding the progress of the detected vehicle M. The signal status of the traffic light TL includes a state indicating progress (blue), a state indicating caution (yellow), a state indicating stop (red), a state indicating right turn possible, a state indicating possible left turn, and a pedestrian can only be crossed. The state etc. which are shown are mentioned. For example, the acquisition unit 16 acquires signal information from the traffic light TL. The acquisition unit 16 may acquire signal information from the traffic light control system. The traffic light control system is a system for controlling the traffic light TL. Alternatively, when a video camera is installed at the intersection P, the acquisition unit 16 may acquire signal information by performing image processing on the moving image data acquired by the video camera. The acquisition unit 16 outputs the signal information to the behavior detection unit 15.

  The behavior detection unit 15 detects signal ignorance of the detected vehicle based on the tracking result and the signal information. Specifically, the behavior detection unit 15 determines whether or not the detected vehicle has entered the intersection P when the traffic light TL indicates a stop, and detects the traffic when the traffic light TL indicates a stop. When it is determined that the vehicle has entered the intersection P, it is determined that the detected vehicle has ignored the signal. Details of the detection processing by the behavior detection unit 15 will be described later. The behavior detection unit 15 outputs an acquisition command for acquiring vehicle information to the vehicle information acquisition device 3 when detecting signal ignorance.

  Next, a detection method performed by the detection device 10 will be described with reference to FIGS. FIG. 5 is a flowchart showing a series of processes of a detection method performed by the detection apparatus. FIG. 6 is a flowchart showing in detail the vehicle behavior detection process of FIG. FIG. 7 is a flowchart showing in detail the analysis processing of FIG. FIG. 8 is a flowchart showing in detail the deletion process of FIG. FIG. 9 is a diagram illustrating an example of the management table. The series of processing shown in FIG. 5 is performed at regular time intervals, for example. In the present embodiment, the traffic light TL shows only stop (red), attention (yellow), and progress (blue).

  First, the acquisition unit 11 acquires point cloud information from the laser radar 2 (step S01). Then, the acquisition unit 11 outputs the acquired point group information to the object detection unit 13. Subsequently, the object detection unit 13 detects a detected object based on the point cloud information (step S02). Specifically, when the object detection unit 13 receives the point cloud information from the acquisition unit 11, the object detection unit 13 obtains the measurement point information of the measurement points located outside the detection area Rd indicated by the area information received from the setting unit 12. Exclude from Then, the object detection unit 13 clusters the remaining point group information and detects the obtained cluster as a single detection object (vehicle, person, etc.). Then, the object detection unit 13 calculates the size and position of the detected object.

  Subsequently, the object detection unit 13 classifies the detected objects (step S03). Specifically, the object detection unit 13 classifies the detected object into a vehicle and a non-vehicle based on the size of the detected object. Then, the object detection unit 13 outputs detection results to the object tracking unit 14 for the detected objects classified as vehicles.

  Subsequently, the object tracking unit 14 tracks the detected object (detected vehicle) (step S04). Specifically, when the object tracking unit 14 receives the detection result from the object detection unit 13, the object tracking unit 14 determines that the detected vehicle detected in the current frame does not correspond to any of the detected vehicles detected in the past frame. In this case, a new vehicle ID is given to the detected vehicle as a new detected vehicle. When the object tracking unit 14 determines that the detected vehicle detected in the current frame corresponds to the detected vehicle detected in the past frame, the object tracking unit 14 sets the vehicle ID assigned to the corresponding detected vehicle in the current frame. It is given to the detected vehicle. Then, the object tracking unit 14 outputs the tracking result (vehicle ID, vehicle position information, and detection time information) to the behavior detection unit 15.

  Subsequently, the acquisition unit 16 acquires signal information (step S05). Then, the acquisition unit 16 outputs the acquired signal information to the behavior detection unit 15.

  Subsequently, the behavior detection unit 15 performs a vehicle behavior detection process (step S06). In the vehicle behavior detection process in step S06, as shown in FIG. 6, when the behavior detection unit 15 receives the tracking result from the object tracking unit 14, the behavior detection unit 15 performs steps S11 to S14 for each vehicle ID included in the tracking result. Process. First, the behavior detection unit 15 updates the management table (step S11). The management table is a table for managing detected vehicles.

  As shown in FIG. 9, the management table manages behavior information in which vehicle IDs, detection time information, and entry information are associated with each other. The detection time information can be expressed in, for example, UNIX (registered trademark) time. The UNIX (registered trademark) time is a time expressed in the number of seconds or milliseconds that have passed since midnight, January 1, 1970. For example, the detection time of the vehicle ID “1001” shown in FIG. 8 is represented by “1516667846000”. The approach information is information indicating whether or not the detected vehicle indicated by the vehicle ID has entered the intersection P. When the entry information is “0”, it indicates that the detected vehicle has not entered the intersection P. When the entry information is “1”, it indicates that the detected vehicle has entered the intersection P. Show.

  If the behavior information including the vehicle ID included in the tracking result does not exist in the management table, the behavior detection unit 15 additionally registers the behavior information including the vehicle ID in the management table. That is, the behavior detection unit 15 adds new behavior information, sets the vehicle ID and detection time information included in the tracking result as the vehicle ID and detection time information of the behavior information, and sets “0” to the entry information. To do. When the behavior information including the vehicle ID included in the tracking result exists in the management table, the behavior detection unit 15 overwrites (updates) the detection time information of the behavior information with the detection time information included in the tracking result.

  Subsequently, the behavior detection unit 15 performs an analysis process (step S12). In the analysis processing in step S12, as shown in FIG. 7, the behavior detection unit 15 first determines whether or not the traffic light TL indicates stop (red) based on the signal information received from the acquisition unit 16. (Step S21). If the behavior detection unit 15 determines that the traffic light TL indicates a stop (step S21; YES), the behavior detection unit 15 determines whether the detected vehicle M has entered the intersection P between the previous frame and the current frame. (Step S22). The behavior detection unit 15 determines whether or not the detected vehicle M has entered the intersection P based on the vehicle position information included in the tracking result. In the behavior detection unit 15, information indicating the range of the intersection P is set in advance.

  An example of the determination method in step S22 will be described. First, the behavior detection unit 15 determines whether or not the detected vehicle M is located in the intersection P based on the vehicle position information included in the tracking result. For example, if the position indicated by the vehicle position information is within the range of the intersection P, the behavior detection unit 15 determines that the detected vehicle M is located within the intersection P, and the position indicated by the vehicle position information is within the range of the intersection P. If it is outside, it is determined that the detected vehicle M is not located in the intersection P. When the behavior detection unit 15 determines that the detected vehicle M is located in the intersection P, the behavior detection unit 15 determines whether the detected vehicle M is from the previous frame to the current frame based on the behavior information including the vehicle ID included in the tracking result. It is determined whether or not the detection vehicle M has entered the intersection P.

  The behavior detection unit 15 determines that the detection vehicle M has entered the intersection P between the previous frame and the current frame when the entry information indicates that the detection vehicle M has not entered the intersection P ( Step S22; YES), “1” is set in the entry information of the behavior information including the vehicle ID. That is, when the entry information indicates that the detected vehicle M has not entered the intersection P, it means that the detected vehicle M has not entered the intersection P until the previous frame. Since the detected vehicle M is located in the intersection P in the current frame, it can be considered that the detected vehicle M has entered the intersection P between the previous frame and the current frame.

  On the other hand, when the detected vehicle M is located outside the intersection P, or when the entry information indicates that the detected vehicle M has entered the intersection P, the detected vehicle M is detected at the intersection P between the previous frame and the current frame. Has not entered. Therefore, when the behavior detection unit 15 determines that the detected vehicle M is located outside the intersection P, or when the entry information indicates that the detected vehicle M has entered the intersection P, the behavior detection unit 15 starts from the previous frame. It is determined that the detected vehicle M has not entered the intersection P before the frame (step S22; NO). The behavior detection unit 15 may determine whether or not the detected vehicle M has entered the intersection P depending on whether or not the detected vehicle M has exceeded the stop line Ls.

  In step S22, when the behavior detection unit 15 determines that the detected vehicle M has entered the intersection P (step S22; YES), the behavior detection unit 15 determines that the detected vehicle M has ignored the signal (step S23). Even if the detection vehicle M enters the intersection P when the traffic light TL indicates progress (blue) or attention (yellow), it does not correspond to signal ignorance. Further, even if the traffic light TL indicates stop (red), if the detected vehicle M has not entered the intersection P, it does not correspond to the signal neglect. For this reason, when the behavior detection unit 15 determines in step S21 that the traffic light TL does not indicate a stop (step S21; NO), and in step S22, the detected vehicle M is detected between the previous frame and the current frame. When it determines with not having entered the intersection P (step S22; NO), the analysis process of step S12 is complete | finished.

  Subsequently, as illustrated in FIG. 6, when the behavior detection unit 15 determines in step S12 that the behavior of the detected vehicle M is signal ignorance (step S13; YES), the behavior detection unit 15 displays the vehicle information of the detected vehicle. In order to acquire, an acquisition command is transmitted to the vehicle information acquisition apparatus 3 with vehicle position information (step S14). At this time, the behavior detection unit 15 may transmit various information related to the violating vehicle to the storage device 5 and the communication device 6. Various types of information may include the detection time, position, speed, and size (dimension) of the detected vehicle (violating vehicle). The various information may include a video movie in a time zone including the detection time. The various types of information may include point cloud information in a time zone including the detection time and a movement trajectory (time-series vehicle position information) of the detected vehicle. Further, the behavior detection unit 15 may transmit an output command for performing a caution, a warning, and an instruction to the violating vehicle to the output device 4. The behavior detection unit 15 does not transmit an acquisition command to the vehicle information acquisition device 3 when it is determined in step S12 that the behavior of the detected vehicle M is not signal ignorance (step S13; NO).

  And the behavior detection part 15 determines whether all the vehicle IDs contained in the tracking result were processed (step S15). If the behavior detection unit 15 determines that all the vehicle IDs have not been processed (step S15; NO), the behavior detection unit 15 executes steps S11 to S15 again for the next vehicle ID. On the other hand, if the behavior detection unit 15 determines that all the vehicle IDs have been processed (step S15; YES), the behavior detection unit 15 performs behavior information deletion processing (step S16).

  In the deletion process in step S16, as shown in FIG. 8, the behavior detection unit 15 performs the processes in steps S31 and S32 for each piece of behavior information registered in the management table. First, the behavior detection unit 15 calculates the difference between the current time and the detection time indicated by the detection time information included in the behavior information, and determines whether the difference is equal to or greater than a preset threshold value. (Step S31). The threshold value is set to a value larger than the average time required for the vehicle to pass through the detection area Rd, for example. When the behavior detection unit 15 determines that the difference is equal to or greater than the threshold (step S31; YES), the behavior detection unit 15 regards that the vehicle indicated by the vehicle ID of the behavior information has left the detection area Rd, and stores the behavior information in the management table. (Step S32). On the other hand, when the behavior detection unit 15 determines that the difference is less than the threshold (step S31; NO), the behavior detection unit 15 does not delete the behavior information from the management table.

  Then, the behavior detection unit 15 determines whether or not all behavior information registered in the management table has been processed (step S33). If the behavior detection unit 15 determines that all the behavior information has not been processed (step S33; NO), the behavior detection unit 15 executes steps S31 to S33 again for the next behavior information. On the other hand, when it is determined that all the behavior information has been processed (step S33; YES), the behavior detection unit 15 ends the vehicle behavior detection process. Thus, a series of processes of the detection method performed by the detection apparatus 10 is completed.

  And the vehicle information acquisition apparatus 3 will acquire vehicle information, if an acquisition command is received with the vehicle position information from the detection apparatus 10 (behavior detection part 15). Specifically, when the vehicle existing at the position indicated by the vehicle position information enters the monitoring range Rb, the vehicle information acquisition device 3 captures the number plate of the vehicle and acquires the number plate image. To do. And the vehicle information acquisition apparatus 3 transmits vehicle information to the memory | storage device 5, makes it memorize | store in the memory | storage device 5, and transmits vehicle information to a public organization, a road management company, etc. via the communication apparatus 6. FIG.

  As described above, in the detection system 1 and the detection apparatus 10, the detection vehicle M is detected using the point cloud information generated by the laser radar 2 irradiating the detection area Rd with laser light, and the detection vehicle M is detected. Is tracked. Based on the tracking result of the detected vehicle M and the signal information indicating the state of the traffic light TL, it is determined whether or not the detected vehicle M has entered the intersection P when the traffic light TL is in a stopped state. When it is determined that the detected vehicle M has entered the intersection P when TL is in a stop state, it is determined that the detected vehicle M has ignored the signal.

  The laser radar 2 is affected by environmental disturbances such as changes in the amount of light depending on the time zone (morning and night), changes in the weather (rainy weather, snowfall, and heavy fog), and strong light such as sunlight and headlights. Hateful. The laser radar 2 can measure the distance between the laser radar 2 and the object with an error of about several centimeters. For this reason, the detection system 1 and the detection apparatus 10 can detect the position of the detection vehicle M with high accuracy.

  For example, in the case where it is detected that the vehicle has entered an intersection due to the laser beam irradiated by the laser rangefinder being blocked by the vehicle, a false detection may occur. For example, a vehicle that turns right from a road that intersects the road where the vehicle exists, and a pedestrian or bicycle that crosses a pedestrian crossing provided across the road where the vehicle exists blocks a laser beam. May be falsely detected as entering the intersection. According to Japanese traffic rules, if a motorcycle is traveling on a road with more than 3 lanes, including a right turn lane, it is necessary to make a two-step right turn at the intersection. Even in such a case, it may be erroneously detected that the vehicle has entered the intersection.

  On the other hand, in the detection system 1 and the detection device 10, the movement track of the detection vehicle M is obtained by tracking the detection vehicle M. The movement trajectory of pedestrians and bicycles crossing the road TR through the pedestrian crossing is different from the movement trajectory of a vehicle entering the intersection P from the road TR. The movement trajectory of a vehicle that makes a right turn from a road that intersects the road TR is also different from the movement trajectory of a vehicle that enters the intersection P from the road TR. For this reason, by tracking the detection vehicle M, it is possible to reduce the possibility that a pedestrian or the like crossing a pedestrian crossing will be erroneously detected as a vehicle entering the intersection P. As a result, it is possible to improve the detection accuracy of ignoring signals.

  The behavior detection unit 15 outputs an acquisition command for causing the vehicle information acquisition device 3 to acquire vehicle information for specifying the vehicle that has ignored the signal. Thereby, the vehicle information for specifying the vehicle which performed signal ignorance is acquired. For this reason, for example, it is possible to give violation points and collect fines for the driver of the vehicle who has ignored the signal. Further, by identifying the vehicle that has ignored the signal, a warning can be given to the driver of the vehicle that has ignored the signal.

  In order to crack down on signal-ignoring vehicles, police officers are deployed at intersections or monitored with surveillance cameras. However, it takes time to monitor as much as real time at one intersection. Moreover, at an intersection with a large amount of traffic, there is a possibility that traffic of other vehicles may be obstructed by detecting a violation vehicle. Therefore, it is desired to efficiently extract a signal-ignoring vehicle and acquire vehicle information. According to the detection system 1 and the detection device 10, it is possible to detect a signal-ignoring vehicle on a constant basis. For this reason, the driver is obliged to comply with the traffic rules when the driver recognizes that the signal ignore is reliably detected. Thereby, the occurrence of an accident can be suppressed.

  Since the laser radar 2 repeatedly irradiates the laser beam to the irradiable range Ra (detection area Rd), the position of the detection vehicle M can be detected over time. Since the laser radar 2 has a wider detection area Rd than a laser distance meter and a camera, it is not necessary to install a plurality of laser radars 2 for a plurality of lanes. The laser radar 2 can measure even in a space with a good line of sight such as an intersection.

  As mentioned above, although embodiment of this indication was described, this invention is not limited to the said embodiment.

  For example, in the above embodiment, the detection device 10 is configured as one device, but may be configured with two or more devices. The detection device 10 may include a laser radar 2. The laser radar 2 may include an information processing device having the function of the detection device 10.

  The intersection is not limited to a crossroad, but may be a T-junction. The road for which signal ignorance is detected may not be a road with one lane (running lane) on one side, but may be a road having a plurality of lanes on one side.

  The arrangement of the laser radar 2 and the vehicle information acquisition device 3 is not limited to the arrangement shown in each drawing. For example, in FIG. 2, the vehicle information acquisition device 3 can photograph the offending vehicle from the front. Since the license plate is also provided at the rear of the vehicle, the vehicle information acquisition device 3 may be provided so that the illegal vehicle can be photographed from the rear. Moreover, both the vehicle information acquisition apparatus 3 which can image | photograph a violation vehicle from the front, and the vehicle information acquisition apparatus 3 which can image | photograph a violation vehicle from back may be provided.

  As the laser radar 2, a type of three-dimensional laser radar that can irradiate laser light in all directions may be used. In this case, since a plurality of detection areas Rd can be monitored by a single laser radar 2, the number of laser radars 2 can be reduced.

  Although the detection system 1 includes a single laser radar 2, the detection system 1 may include a plurality of laser radars 2. In this case, the detection apparatus 10 may combine and use the point cloud information generated by the plurality of laser radars 2. In this configuration, since the blind spots are reduced, it is possible to improve the certainty of detection.

  The vehicle information acquisition device 3 may be a video camera having high resolution instead of the automatic speed violation control device. The vehicle information acquisition device 3 may further include a flash device (strobe device) in order to cope with nighttime. The vehicle information acquisition device 3 may acquire vehicle information (information described on the number plate) by wireless communication between the communication device 6 and the vehicle-mounted device, for example.

  The vehicle information acquired by the vehicle information acquisition device 3 is not limited to a license plate image. The vehicle information may be a face photo of the driver of the vehicle. The vehicle information may be electronic information such as driver's license information.

  When a gate (breaker) is installed at the destination of the violating vehicle, the detection apparatus 10 may control the gate to be blocked in order to prevent the violating vehicle from physically traveling. In this case, the traveling of the violating vehicle can be obstructed.

  When the detection system 1 and the detection apparatus 10 can intervene in the traveling control system of the violating vehicle, the violating vehicle may be safely stopped.

  In the above embodiment, the behavior detection unit 15 detects the detected vehicle M that has entered the intersection P as the signal ignoring vehicle while the traffic light TL indicates the stop. Not limited to. For example, the behavior detection unit 15 enters the intersection P in an all-stop state (all red state) in which all the traffic lights TL provided at the intersection P indicate a stop in order to detect only more malicious signal ignorance. The detected vehicle M may not be detected as a signal ignoring vehicle. The behavior detection unit 15 may detect the detected vehicle M that has entered the intersection P in a fully stopped state as a signal ignoring vehicle.

  If the vehicle enters the intersection P while the traffic light TL indicates caution (yellow), the signal may be ignored. That is, when the traffic light TL is a yellow signal, the vehicle may have to stop. For this reason, the behavior detection unit 15 may detect the detected vehicle M that has entered the intersection P as a signal-ignoring vehicle while the traffic light TL indicates the attention. In other words, the state in which the traffic light indicates stop may include the state in which the traffic light indicates attention.

  The behavior detection unit 15 may classify the detection vehicle M according to the signal state of the traffic light TL indicated by the signal information when the detection vehicle M enters the intersection P. For example, when the traffic signal TL indicates stop (red), the behavior detection unit 15 detects the vehicle M that has entered the intersection P, and all the traffic signals TL provided at the intersection P indicate stop (red). Each detected vehicle M is classified into a detected vehicle M that has entered the intersection P in (all stopped state) and a detected vehicle M that has entered the intersection P when the traffic light TL indicates caution (yellow). The behavior detection unit 15 may include classification information indicating the classification of the detected vehicle M in various information related to the violating vehicle. The danger varies depending on the signal state when the detection vehicle M enters the intersection P. Therefore, by classifying the detection vehicle M according to the signal state, a warning or the like can be given to the driver according to the danger.

  As shown in FIGS. 10 to 15, the traffic light TL may indicate that a specific direction can pass in addition to travel (blue), attention (yellow), and stop (red). FIG. 10 is a diagram for explaining a modified example of signal ignore detection. FIG. 11 is a flowchart showing in detail the analysis processing in the modification shown in FIG. FIG. 12 is a diagram showing an example of the management table in the modification shown in FIG. FIG. 13 is a diagram for explaining the passage of the detection line. FIG. 14 is a diagram for explaining signal ignorance in the modification shown in FIG. FIG. 15 is a diagram for explaining the progress in a specific direction in the modification shown in FIG.

  In the example shown in FIG. 10, roads TR <b> 1 to TR <b> 4 extend from the intersection P in all directions. For example, the road TR1 has a plurality of (here, three) lanes that enter the intersection P. The traffic light TL may indicate that a right turn is possible. At such an intersection P, even if the traffic light TL indicates a stop, the vehicle can turn right if the traffic light TL indicates that a right turn is possible.

  The detection method of the present modification is mainly different from the detection method of the above embodiment in the analysis process (step S12). As shown in FIG. 11, in the analysis process (step S12) of this modification, the behavior detection unit 15 first stops the traffic light TL based on the signal information received from the acquisition unit 16, as in step S21. It is determined whether or not (red) is indicated (step S41). If the behavior detection unit 15 determines that the traffic light TL indicates a stop (step S41; YES) as in step S22, the detection vehicle M enters the intersection P between the previous frame and the current frame. Is determined (step S42). The behavior detection unit 15 determines whether or not the detected vehicle M has entered the intersection P based on the vehicle position information included in the tracking result.

  In step S42, when it is determined that the detected vehicle M has entered the intersection P between the previous frame and the current frame (step S42; YES), the behavior detection unit 15 records the entry at the time of a red signal ( Step S43). As shown in FIG. 12, in the management table of the present modification, the behavior information further includes red signal entry information in addition to the vehicle ID, the detection time information, and the entry information. The red signal approach information is information indicating whether or not the detected vehicle indicated by the vehicle ID has entered the intersection P at the time of a red signal. When the red signal approach information is “0”, it indicates that the detected vehicle has not entered the intersection P at the time of the red signal, and when the red signal approach information is “1”, the detected vehicle is It indicates that the vehicle has entered the intersection P at the time of red light. In step S <b> 43, the behavior detection unit 15 sets “1” in the red signal entry information of the behavior information corresponding to the vehicle ID of the detected vehicle M. And the behavior detection part 15 performs step S44.

  Even if the detection vehicle M enters the intersection P when the traffic light TL indicates progress (blue) or attention (yellow), it does not correspond to signal ignorance. Further, even if the traffic light TL indicates stop (red), if the detected vehicle M does not enter the intersection P, it does not correspond to the signal neglect. Therefore, when the behavior detection unit 15 determines in step S41 that the traffic light TL does not indicate a stop (step S41; NO), and in step S42, the detected vehicle M is detected between the previous frame and the current frame. If it is determined that the vehicle has not entered the intersection P (step S42; NO), step S44 is executed.

  Subsequently, the behavior detection unit 15 determines whether or not the entry at the time of a red signal is recorded (step S44). Specifically, the behavior detection unit 15 determines whether or not an approach at the time of a red signal is recorded based on the red signal approach information of the behavior information including the vehicle ID included in the tracking result. When the behavior detection unit 15 determines that the red signal approach information is “0” and the approach at the time of the red signal is not recorded (step S44; NO), the detected vehicle M is ignoring the signal. It is determined that there is not, and the analysis processing in step S12 is terminated.

  On the other hand, when the behavior detection unit 15 determines in step S44 that the red signal approach information is “1” and the approach at the time of the red signal is recorded (step S44; YES), the detected vehicle M is It is determined whether or not the detection lines Lout1 to Lout3 have been exceeded (step S45). Detection lines Lout1 to Lout3 are set at boundaries between the intersection P and roads TR2 to TR4 (a plurality of roads) extending from the intersection P. The detection lines Lout1 to Lout3 are virtual lines for detecting that the detection vehicle M has left the intersection P. The detection line Lout1 is provided at the boundary between the intersection P and the road TR2. The detection line Lout2 is provided at the boundary between the intersection P and the road TR3. The detection line Lout3 is provided at the boundary between the intersection P and the road TR4.

  The behavior detection unit 15 determines whether or not the detection line has been exceeded using the position of the detected vehicle M indicated by the vehicle position information in the previous (time t-1) frame and the current (time t) frame. Specifically, as illustrated in FIG. 13, the behavior detection unit 15 is positioned on one side of two spaces where the position p (t−1) of the detection vehicle M in the previous frame is separated by the detection line L. Then, when the position p (t) of the detection vehicle M in the current frame is located on the other side, it is determined that the detection vehicle M has exceeded the detection line L. In other words, when the vehicle position information indicates the barycentric position of the detected vehicle M, the behavior detection unit 15 when the line segment Lg connecting the barycentric positions of the detected vehicle M between two consecutive frames intersects the detection line L. Determines that the detection vehicle M has exceeded the detection line L.

  In step S45, when the behavior detection unit 15 determines that the detected vehicle M has exceeded the detection line (step S45; YES), the behavior detection unit 15 specifies the traveling direction of the detected vehicle M (step S46). The fact that the detection vehicle M has exceeded any of the detection lines Lout1 to Lout3 means that the detection vehicle M has left the intersection P. For this reason, the behavior detection part 15 specifies the advancing direction of the detection vehicle M according to which detection line of the detection lines Lout1-Lout3 the detection vehicle M exceeded. If the behavior detection unit 15 determines that the detection vehicle M has exceeded the detection line Lout1, the behavior detection unit 15 determines that the detection vehicle M has turned left. The behavior detection unit 15 determines that the detection vehicle M has traveled straight when it is determined that the detection vehicle M has exceeded the detection line Lout2. The behavior detection unit 15 determines that the detection vehicle M has turned right when it is determined that the detection vehicle M has exceeded the detection line Lout3.

  And the behavior detection part 15 determines whether the advancing direction of the detection vehicle M is a passable direction based on signal information (step S47). For example, as shown in FIGS. 14 and 15, it is assumed that the traffic light TL indicates a stop (red) and a right turn is possible. At this time, in the example shown in FIG. 14, the detection vehicle M is traveling straight. Therefore, the behavior detection unit 15 determines that the traveling direction (straight forward) of the detected vehicle M is not a passable direction (right turn) (step S47; NO), and determines that the detected vehicle M has ignored the signal (step S47). S48). And the behavior detection part 15 complete | finishes the analysis process of step S12.

  On the other hand, in the example shown in FIG. 15, the detection vehicle M is turning right. Therefore, the behavior detection unit 15 determines that the traveling direction (right turn) of the detected vehicle M is a passable direction (right turn) (step S47; YES), and determines that the detected vehicle M does not ignore the signal. To do. And the behavior detection part 15 complete | finishes the analysis process of step S12.

  In step S45, when the behavior detection unit 15 determines that the detection vehicle M does not exceed any detection line (step S45; NO), the behavior detection unit 15 determines that the detection vehicle M remains in the intersection P ( Step S49) and the analysis process of step S12 are terminated. In addition, when the detected vehicle M that has entered the intersection P at the time of a red signal stays at the intersection P for a certain time, the detected vehicle M may be a violation vehicle.

  As described above, there is a traffic light TL indicating that a specific direction can be passed while indicating a stop. At the intersection P where such a traffic light TL is installed, each vehicle can travel in a direction in which it can pass even at a red light. In such a situation, in the above modification, the behavior detection unit 15 signals that the traffic light TL indicates the direction in which the vehicle can travel, so that a vehicle traveling in the direction in which the vehicle can travel is not erroneously detected as ignoring the signal. When the information further indicates, it is determined whether or not the detected vehicle M has ignored the signal by comparing the traveling direction of the detected vehicle M with the direction in which the vehicle can pass. Thereby, when the detection vehicle M travels in a passable direction, it is possible to prevent the detection vehicle M from being determined to have ignored the signal. As a result, it is possible to further improve the detection accuracy of ignoring signals.

  The behavior detection unit 15 specifies the traveling direction of the detection vehicle M when the detection vehicle M exceeds any of the detection lines Lout1 to Lout3. When the detection vehicle M that has entered the intersection P leaves the intersection P, the detection vehicle M always exceeds any of the detection lines Lout1 to Lout3. For this reason, by using the detection lines Lout1 to Lout3, it is possible to reliably detect the exit from the intersection P regardless of the speed of the detection vehicle M, and to improve the identification accuracy of the traveling direction of the detection vehicle M. It becomes possible. As a result, it is possible to further improve the detection accuracy of ignoring signals.

  In the above embodiment and the modification, the signal ignorance of the vehicle is the detection target, but the detection target is not limited to the signal ignorance of the vehicle. For example, pedestrians (people) and bicycle signal ignorance may be detected. In this case, the acquisition unit 16 may acquire signal information indicating the signal state of the pedestrian traffic light. Moreover, the detection system 1 may include an object information acquisition device that acquires object information for identifying an object for which signal ignorance has been performed. Examples of the object information acquisition device include an imaging device. The behavior detection unit 15 outputs an acquisition command to the object information acquisition device together with the position information of the detected object in order to acquire object information for specifying the object for which the signal is ignored. When the object information acquisition device receives the acquisition command together with the position information from the detection device 10, for example, the object information acquisition device performs imaging so as to include an object existing at the position indicated by the position information, and acquires an image of the object. For example, when the object to be detected is a pedestrian, an image including the pedestrian's face is acquired.

1 Detection System 2 Laser Radar 3 Vehicle Information Acquisition Device (External Device)
4 Output device 5 Storage device 6 Communication device 10 Detection device 11 Acquisition unit (first acquisition unit)
12 setting unit 13 object detection unit 14 object tracking unit 15 behavior detection unit 16 acquisition unit (second acquisition unit)
Lout1, Lout2, Lout3 Detection line M Detection vehicle P Intersection Rd Detection area TL Traffic light

Claims (6)

A detection device that detects signal ignorance using a laser radar that generates point cloud information by irradiating a detection area set at an intersection provided with a traffic light with a laser beam and receiving reflected light of the laser beam. There,
A first acquisition unit for acquiring the point cloud information;
An object detection unit for detecting an object based on the point cloud information;
An object tracking unit that tracks the object detected by the object detection unit and generates a tracking result;
A second acquisition unit that acquires signal information indicating a signal state of a traffic light that gives an instruction regarding the progress of the object;
Based on the tracking result and the signal information, a behavior detection unit that detects signal ignorance of the object;
With
The behavior detection unit determines whether or not the object has entered the intersection when the traffic signal indicates a stop, and the object enters the intersection when the traffic signal indicates a stop. A detection device that determines that the object has ignored the signal when it is determined that the signal has been ignored.   The behavior detection unit is configured to compare the traveling direction of the object and the passable direction when the signal information further indicates that the traffic signal indicates a passable direction. The detection device according to claim 1, wherein it is determined whether or not the object has ignored the signal. Detection lines for detecting that the object has left the intersection are set at boundaries between the intersection and a plurality of roads extending from the intersection,
The detection device according to claim 2, wherein the behavior detection unit specifies the traveling direction when the object exceeds the detection line.   The said behavior detection part classifies the said object according to the said signal state shown by the said signal information when the said object approached the said intersection, The description in any one of Claims 1-3 Detection device.   The detection device according to any one of claims 1 to 4, wherein the behavior detection unit outputs an acquisition command that causes an external device to acquire object information for specifying the object for which the signal is ignored. The detection device according to any one of claims 1 to 5,
The laser radar;
A detection system comprising:

Images