Traffic signal control device

Abstract

To make it possible to fully enhance the precision in traffic light control for improving a state involving traffic jam on a right turn lane in the event that the traffic jam has occurred on the right turn lane, and fully enhance the precision in traffic light control for improving a state involving pedestrians waiting for traverse or congestion of vehicles trying to turn left in the event that the pedestrians waiting for traverse or congestion of vehicles trying to turn left has occurred.SOLUTION: A traffic light controller is configured to: acquire, by performing image analysis processing on a camera image, traffic state information including information concerning a position of a trailing vehicle in a string of vehicles congested in a monitoring area; and perform traffic light control based on the traffic state information. The traffic light controller is further configured to: acquire, by performing the image analysis processing on the camera image, information including information concerning a passing state of vehicles intending to turn left in a monitoring area targeting the vehicles intending to turn left, information concerning a passing state of traversing pedestrians in a motoring area targeting the traversing pedestrians, and information concerning a standby state of pedestrians who wait for traversing in a monitoring area targeting the pedestrians who wait for traversing; and perform traffic light control based on the traffic state information.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to a traffic signal control device that controls traffic signals installed at road intersections. [Background technology]

[0002] In a traffic control system (traffic signal control system), a central control device installed in a traffic control center collects traffic condition information using vehicle detectors installed on roads, generates signal control information based on that traffic condition information to facilitate the passage of vehicles and pedestrians at intersections, and controls the traffic signals installed at the intersections based on that signal control information.

[0003] Such traffic control systems can only collect traffic information at the locations where vehicle detectors are installed, which limits their ability to provide optimal signal control. For example, when a traffic jam occurs in a right-turn lane due to a high volume of right-turning vehicles, the location of the end of the jammed vehicle line, i.e., the length of the jam, is unknown, making it impossible to sufficiently improve the accuracy of signal control to alleviate the congestion in the right-turn lane. Also, when a high volume of pedestrian traffic causes pedestrians to wait to cross or left-turning vehicles to accumulate, the number of pedestrians passing (the number of people crossing), the number of pedestrians waiting (the number of people waiting to cross), and the number and average speed of left-turning vehicles are unknown, making it impossible to sufficiently improve the accuracy of signal control to alleviate the congestion of pedestrians waiting to cross or left-turning vehicles to accumulate.

[0004] In relation to such problems, a technology has been known in the past in which an image is acquired from a camera installed to capture the area around an intersection, and the captured image is subjected to image analysis processing to detect vehicles (moving objects) entering the intersection, and based on the detection results, the volume of traffic entering the intersection is measured, and based on the measurement results, signal control information is generated (see Patent Document 1). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent Publication No. 2021-82362 Summary of the Invention [Problem to be solved by the invention]

[0006] Conventional technology can detect moving objects entering an intersection. However, when a traffic jam occurs in the right-turn lane, no consideration is given to measuring the end position of the congested vehicle line, i.e., the length of the traffic jam, which means that the accuracy of signal control to alleviate the congestion in the right-turn lane cannot be sufficiently improved. Furthermore, when pedestrians are waiting to cross or left-turning vehicles are stuck, no consideration is given to measuring the number of pedestrians passing or waiting, or the number and average speed of left-turning vehicles, which means that the accuracy of signal control to alleviate the congestion of pedestrians waiting to cross or left-turning vehicles cannot be sufficiently improved.

[0007] Therefore, the main object of the present invention is to provide a traffic signal control device that can sufficiently improve the accuracy of signal control to alleviate congestion in the right-turn lane when congestion occurs in the right-turn lane, and that can sufficiently improve the accuracy of signal control to alleviate congestion of pedestrians waiting to cross or left-turning vehicles when congestion occurs in the right-turn lane. [Means for solving the problem]

[0008] The traffic signal control device of the present invention is a traffic signal control device having a processor that executes processing to generate signal control information for controlling a traffic signal installed at an intersection based on traffic condition information around the intersection, and the processor acquires detection results of a sensor that detects an object on the road, and performs an analysis process on the detection results of the sensor. Exchange Get status information hand , and generates signal control information based on the traffic situation information. In generating the signal control information, the end position of the congested vehicle queue in the right-turn lane is acquired as the traffic condition information based on the occupancy status of objects on the road surface within a detection frame corresponding to each observation position for each of a plurality of observation positions set within the monitoring area, and the signal time for allowing right-turning vehicles to proceed is adjusted in stages depending on which observation position the end position is, and the average speed of left-turning vehicles and the number of crossing pedestrians are acquired as the traffic condition information based on the traffic status of left-turning vehicles and the traffic status of crossing pedestrians, and at least one of the signal time for allowing left-turning vehicles to proceed and the signal time for allowing pedestrians to cross is adjusted depending on the combination of the average speed of the left-turning vehicles and the number of crossing pedestrians. The configuration will be as follows. [Effects of the Invention]

[0010] According to the present invention, when a traffic jam occurs in a target lane, information on the end position of the queue of vehicles in the traffic jam, i.e., information on the length of the traffic jam, is obtained, thereby sufficiently improving the accuracy of signal control to alleviate the traffic jam in the target lane.Furthermore, when pedestrians are waiting to cross or left-turning vehicles are stuck, information on the traffic status of left-turning vehicles, information on the traffic status of crossing pedestrians, and information on the waiting status of pedestrians waiting to cross is obtained, thereby sufficiently improving the accuracy of signal control to alleviate the pedestrians waiting to cross or the stuck status of left-turning vehicles. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a diagram illustrating the overall configuration of a signal control system according to an embodiment of the present invention. [Figure 2] An explanatory diagram showing the situation of obstacles for right-turning vehicles [Figure 3] An explanatory diagram showing the situation of an obstacle for a vehicle traveling straight [Figure 4] An explanatory diagram showing the situation regarding left-turning vehicles and crossing pedestrians [Figure 5] An explanatory diagram showing the installation of cameras at intersections [Figure 6] An explanatory diagram showing a camera image of the area around an intersection. [Figure 7] FIG. 1 is an explanatory diagram showing the contents of traffic condition information acquired by image analysis processing performed by a signal control server. [Figure 8] Block diagram showing the schematic configuration of the signal control server [Figure 9] Flow diagram showing the processing steps performed by the signal control server [Figure 10] Flow diagram showing the procedure for right-turn improvement processing performed by the signal control server [Figure 11] Flow diagram showing the procedure for improving straight-line driving performed by the signal control server [Figure 12] Flow diagram showing the procedure for left-turn crossing improvement processing performed by the signal control server [Figure 13] An explanatory diagram showing the signal control status DETAILED DESCRIPTION OF THE INVENTION

[0012] The first invention made to solve the above problems is a traffic signal control device having a processor that executes processing to generate signal control information for controlling a traffic signal installed at an intersection based on traffic condition information around the intersection, the processor acquiring detection results of a sensor that detects an object on the road, and analyzing the detection results of the sensor. Exchange Get status information hand , and generates signal control information based on the traffic situation information. In generating the signal control information, the end position of the congested vehicle queue in the right-turn lane is acquired as the traffic condition information based on the occupancy status of objects on the road surface within a detection frame corresponding to each observation position for each of a plurality of observation positions set within the monitoring area, and the signal time for allowing right-turning vehicles to proceed is adjusted in stages depending on which observation position the end position is, and the average speed of left-turning vehicles and the number of crossing pedestrians are acquired as the traffic condition information based on the traffic status of left-turning vehicles and the traffic status of crossing pedestrians, and at least one of the signal time for allowing left-turning vehicles to proceed and the signal time for allowing pedestrians to cross is adjusted depending on the combination of the average speed of the left-turning vehicles and the number of crossing pedestrians. The configuration will be as follows.

[0013] This allows for the acquisition of information on the end position of the queue of vehicles in a traffic jam, i.e., the length of the traffic jam, when a traffic jam occurs in the target lane, thereby enabling the accuracy of signal control to improve the traffic jam in the target lane to be sufficiently improved. Note that the sensor may be a camera or other sensor, such as a lidar or radar.

[0016] Also, 2 The invention is configured such that the processor acquires information regarding the speed of straight-moving vehicles based on the detection results of the sensor regarding the monitoring area targeting the straight-moving lane, and adjusts the signal time for allowing straight-moving vehicles to pass based on that information.

[0017] According to this, when the passage of straight-moving vehicles is not smooth, the time of the signal allowing the passage of straight-moving vehicles can be adjusted to facilitate the passage of straight-moving vehicles. In this case, the average speed of the straight-moving vehicles may be acquired, and when the average speed of the straight-moving vehicles is equal to or less than a predetermined threshold, the time of the signal allowing the passage of straight-moving vehicles may be extended.

[0024] Also, 3In the invention, the processor is configured to generate signal control information that facilitates the passage of vehicles and pedestrians from current traffic condition information using a signal control model constructed by learning using traffic condition information at each point in time in the past, signal control information used at that point in time, and the control results at that point in time.

[0025] This allows for more appropriate signal control to alleviate congestion in the right-turn lane, as well as signal control to alleviate the congestion of pedestrians waiting to cross and left-turning vehicles.

[0026] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0027] FIG. 1 is a diagram showing the overall configuration of a traffic signal control system according to this embodiment.

[0028] A traffic signal control system aims to facilitate the smooth passage of moving objects (vehicles, pedestrians, and bicycles) at intersections. The traffic signal system comprises a camera 1, a signal control server 2 (traffic signal control device), and a traffic signal 3. The camera 1 and the traffic signal 3 are installed at each intersection and connected to the signal control server 2 via a network.

[0029] The camera 1 captures an image of the area around the intersection and transmits the captured image (camera image) to the traffic light control server 2.

[0030] The signal control server 2 constitutes the central device of the traffic control system. The signal control server 2 generates traffic condition information regarding the traffic conditions around the intersection by performing image analysis processing on the camera images, and generates signal control information for controlling the traffic signals 3 based on the traffic condition information. The signal control server 2 transmits the signal control information to the traffic signals 3.

[0031] The traffic signal 3 is composed of a signal controller 31, a vehicle signal light 32, and a pedestrian signal light 33. The signal controller 31 controls the vehicle signal light 32 and the pedestrian signal light 33 based on signal control information (command information) received from the signal control server 2. The signal controller 31 transmits signal operation information (performance information) to the signal control server 2.

[0032] In this embodiment, the camera 1 is provided as a sensor for detecting objects on the road, but the sensor for detecting objects on the road is not limited to the camera 1. For example, the sensor for detecting objects on the road may be a lidar or radar. In addition, the traffic light control server 2 performs image analysis processing on the camera image as the detection result of the camera 1, but similar analysis processing is performed on the detection result of a sensor other than the camera 1.

[0033] Furthermore, in this embodiment, a case where a vehicle travels on the left side of the road (left-hand traffic) will be described, but a case where a vehicle travels on the right side of the road (right-hand traffic) is also possible. In this case, right turns and left turns are reversed. That is, a right turn (a lane change that crosses an oncoming vehicle) when driving on the left side of the road is equivalent to a left turn when driving on the right side of the road, and a left turn (a lane change that does not cross an oncoming vehicle) when driving on the left side of the road is equivalent to a right turn when driving on the right side of the road.

[0034] Next, we will explain the obstacles to vehicle traffic and pedestrians crossing at intersections. Fig. 2 is an explanatory diagram showing the obstacles to vehicles turning right. Fig. 3 is an explanatory diagram showing the obstacles to vehicles going straight. Fig. 4 is an explanatory diagram showing the obstacles to vehicles turning left and pedestrians crossing.

[0035] 2, at an intersection, if there are too many right-turning vehicles, right-turning vehicles cannot pass through the intersection when the right-turn arrow on the vehicle signal lamp 32 is green (right-turn green arrow time), causing vehicles to back up in the right-turn lane and congestion in the right-turn lane. Also, vehicles that cannot enter the right-turn lane due to congestion in the right-turn lane block the straight-going left-turn lane, preventing straight-going and left-turning vehicles from passing through the intersection, and congestion in the right-turn lane causes congestion in the straight-going left-turn lane.

[0036] Here, for example, if a vehicle detector is installed in the right-turn lane, it will be possible to determine that a traffic jam has occurred in the right-turn lane, but it will not be possible to determine the status of right-turning vehicles in the right-turn lane, particularly the position of the end of the queue of congested vehicles, i.e., the length of the congestion.

[0037] Therefore, in this embodiment, image analysis processing is performed on camera images taken around the intersection by camera 1 to obtain information regarding the congestion status of right-turning vehicles, such as the position of the end of a traffic jam, and information regarding the traffic status of right-turning vehicles, such as the number and average speed of right-turning vehicles, and based on this information, signal control is performed to improve the problems related to right-turning vehicles, in particular signal control (e.g., MODERATO control) to reduce congestion of right-turning vehicles.

[0038] Also, as shown in Figure 3, at an intersection, if there are too many straight-moving vehicles, they cannot pass through the intersection during the green light period of the vehicle signal light 32 (vehicle green period), causing vehicles to stagnate in the straight-moving left-turn lane and resulting in congestion in the straight-moving left-turn lane.

[0039] Therefore, in this embodiment, image analysis processing is performed on camera images taken around the intersection by camera 1 to obtain information on the congestion status of straight-moving vehicles, such as the position of the end of a traffic jam, and information on the traffic status of straight-moving vehicles, such as the number and average speed of straight-moving vehicles, and based on this information, signal control is performed to alleviate obstacles related to straight-moving vehicles.

[0040] As shown in Figure 4, at intersections, pedestrians crossing the crosswalk (crossing pedestrians) and left-turning vehicles intersect, but because crossing pedestrians have priority over left-turning vehicles, left-turning vehicles can cross the crosswalk when the crossing pedestrians stop. However, if there are many pedestrians crossing the crosswalk and the crossing continues, left-turning vehicles cannot pass through the intersection, so vehicles stop in the straight-through left-turn lane. As a result, straight-through vehicles cannot pass through the intersection either, causing congestion in the straight-through left-turn lane. Furthermore, if there are too many pedestrians crossing the crosswalk, they cannot all cross the crosswalk when the pedestrian signal light 33 is green (pedestrian green time), and pedestrians waiting in front of the crosswalk (pedestrians waiting to cross) will be stuck.

[0041] For example, if a vehicle detector is installed in the straight-ahead left-turn lane, it can tell that congestion has occurred in the straight-ahead left-turn lane, but it cannot tell that the cause of the congestion is a large number of pedestrians crossing the street. Furthermore, it is also unclear what the situation is with the pedestrians crossing the street, or how many left-turning vehicles are stuck there. It is possible to infer that the cause is a large number of pedestrians crossing the street and control the traffic light to extend the green time for pedestrians, but if the level of congestion of pedestrians waiting to cross the street is unknown, it is not possible to set the green time for pedestrians to an optimal value.

[0042] Therefore, in this embodiment, image analysis processing is performed on camera images taken around the intersection by camera 1 to obtain information regarding the traffic conditions of left-turning vehicles, such as the number of left-turning vehicles passing and their average speed, information regarding the crossing conditions of pedestrians, such as the number of pedestrians passing and their average speed, and information regarding the congestion of pedestrians waiting to cross, such as the number of pedestrians waiting (number of pedestrians waiting to cross), and based on this information, signal control (e.g., matrix control) is performed to alleviate obstacles related to left-turning vehicles and crossing pedestrians.

[0043] Next, we will explain the camera image captured by camera 1 and the monitoring area and observation position set on the camera image. Fig. 5 is an explanatory diagram showing the installation status of camera 1 at an intersection. Fig. 6 is an explanatory diagram showing the camera image captured by camera 1 around the intersection.

[0044] As shown in Fig. 5, camera 1 is installed at an intersection. As shown in Fig. 6, camera 1 captures camera images of the area around the intersection, and signal control server 2 acquires traffic condition information by performing image analysis processing on the camera images. At this time, since monitoring areas are set in the camera images, traffic condition information is acquired for each monitoring area.

[0045] In the example shown in Figure 6, a first monitoring area (straight-through left-turn lane area), a second monitoring area (right-turn lane area), a third monitoring area (left-turn waiting area), a fourth monitoring area (pedestrian crossing area), and a fifth monitoring area (crosswalk waiting area) are set on the camera image.

[0046] The second monitoring area (right-turn lane area) is set in the right-turn lane area within the camera image. The second monitoring area is used to acquire traffic condition information related to the accumulation (congestion) status of right-turning vehicles. Specifically, multiple detection frames corresponding to multiple observation positions are set in the second monitoring area. In the example shown in FIG. 6, rectangular detection frames corresponding to the first to fifth observation positions are set in the second monitoring area. Based on the occupancy status of objects (vehicles) within these multiple detection frames, the tail position of the congested vehicle queue in the right-turn lane, i.e., the congestion length (distance to the tail vehicle), is acquired. Furthermore, based on the occupancy status of objects (vehicles) within the multiple detection frames, the number and average speed of right-turning vehicles are acquired.

[0047] At this time, an image within the detection frame at each observation position is extracted from the camera image, and based on that image, it is determined whether or not any object (usually a vehicle) exists on the road surface within the detection frame. Based on the presence or absence of an object within the detection frame, the occupancy status of the object (vehicle), specifically the space occupancy rate, at each observation position is obtained. Then, based on the change in the space occupancy rate at each observation position, the end of the congested vehicle queue in the right-turn lane is detected, and the traffic status (stagnation status) of vehicles on the right-turn lane is estimated.

[0048] Furthermore, the system can accurately determine whether an object is present within the detection frame, even during the dark hours of the night. This allows the system to accurately detect the end of a traffic jam and estimate the traffic conditions of vehicles in the right-turn lane, even during the dark hours of the night.

[0049] The first monitoring area (straight-ahead left-turn lane area) is set in the area of ​​the straight-ahead left-turn lane in the camera image. The first monitoring area is used to acquire traffic condition information related to the traffic conditions of vehicles on the straight-ahead left-turn lane. Specifically, similar to the second monitoring area, multiple detection frames corresponding to multiple observation positions are set, and the congestion length (distance to the tail vehicle) in the straight-ahead left-turn lane is acquired based on the occupancy status of objects within the detection frame at each observation position, and the number and average speed of vehicles passing on the straight-ahead left-turn lane are also acquired.

[0050] The position of the detection frame (observation position) may be set based on the maximum congestion length expected in the target lane or the maximum congestion length that can be improved by signal control. Also, if the distance from camera 1 to the subject in the camera image increases and exceeds the detection limit, it becomes impossible to ensure detection accuracy regarding the occupancy status of the object within the detection frame, so the position of the detection frame may be set based on the detection limit of the camera image.

[0051] Furthermore, in the process of detecting the tail of a queue of vehicles in a traffic jam in the right-turn lane based on the occupancy status of objects within the detection frame, the size of the vehicles (small vehicles, large vehicles, etc.) may be determined. This allows the number and sizes of vehicles in the queue of vehicles to be identified, and the length of the traffic jam to be estimated more accurately based on this information.

[0052] The third monitoring area (left-turn waiting area) is set up where left-turning vehicles wait in front of the crosswalk. The third monitoring area is used to obtain traffic condition information regarding the congestion of left-turning vehicles. Specifically, vehicle detection is performed using object recognition technology. The number of left-turning vehicles is obtained by counting the detected vehicles. In addition, the average speed of left-turning vehicles is obtained by measuring the speed of each detected vehicle. In addition, by performing a process to track detected vehicles, it is possible to avoid confusing the straight lane of the intersecting road with left-turning vehicles.

[0053] The fourth monitoring area (pedestrian crossing area) is set in the crosswalk area within the camera image. The fourth monitoring area is used to obtain traffic condition information regarding pedestrians crossing the street. Specifically, person detection is performed using object recognition technology. The number of pedestrians passing by is obtained by counting the detected people. The average pedestrian speed is obtained by measuring the speed of each detected person. Bicycle detection is also performed in the fourth monitoring area using object recognition technology. The number of bicycles passing by is obtained by counting the detected bicycles. The average bicycle speed is obtained by measuring the speed of each detected bicycle.

[0054] The fifth monitoring area (waiting to cross) is set in the area of ​​the sidewalk adjacent to the crosswalk in the camera image. The fifth monitoring area is used to obtain traffic condition information related to the status of pedestrians waiting to cross. Specifically, people are detected using object recognition technology, and the number of pedestrians waiting is obtained by counting the detected people. In addition, a process to detect bicycles is performed within the fifth monitoring area, and the number of bicycles waiting is obtained by counting the detected bicycles.

[0055] In the third monitoring area, vehicle detection is performed using object recognition technology, and in the fourth and fifth monitoring areas, person detection and bicycle detection are performed using object recognition technology. However, as with the second monitoring area, traffic condition information may be obtained based on the occupancy status of objects within detection frames set for each of multiple observation positions.

[0056] The third, fourth, and fifth monitoring areas are close to camera 1, so vehicles, people, and bicycles are captured large, allowing for highly accurate object recognition. On the other hand, the first and second monitoring areas are far from camera 1, making highly accurate object recognition difficult. Also, in the first and second monitoring areas, the vehicle at the front of the convoy can be detected by object recognition, but the other vehicles are difficult to detect by object recognition because they are obscured by the vehicle in front and only partially captured. For this reason, processing based on the occupancy status of objects within detection frames set for each of multiple observation positions is highly effective.

[0057] Note that vehicle, person, and bicycle detection using object recognition technology may use an image recognition model (machine learning model) constructed by machine learning such as deep learning. In this case, an image of the monitoring area (area image) extracted from a camera image is input to the image recognition model, and the image recognition model outputs a detection result.

[0058] The position of the monitoring area and the observation position (position of the detection frame) on the camera image are set by the administrator. For example, the camera image may be displayed on a terminal operated by the administrator, and the administrator may specify the position of the monitoring area and the observation position (position of the detection frame) on the camera image.

[0059] Furthermore, camera 1 is installed to capture images of the periphery of the intersection, but the shooting direction and shooting range (angle of view) of camera 1 may be changed as necessary. In this case, the position of the monitored area on the camera image changes, so the position of the monitored area on the camera image and the observation position (position of the detection frame) may be reset by an administrator's operation in accordance with the change in the shooting direction and shooting range of camera 1.

[0060] Next, a description will be given of traffic condition information acquired by the image analysis process performed by the signal control server 2. Fig. 7 is an explanatory diagram showing the contents of the traffic condition information.

[0061] The signal control server 2 performs image analysis processing on the camera images, and by this image analysis processing, traffic condition information regarding the traffic conditions of vehicles and pedestrians within the intersection is obtained.

[0062] In the example shown in FIG. 7, the following information is acquired: date and time, number of areas, lane number, and distance to the tail vehicle (length of the queue of vehicles in a traffic jam). Furthermore, the following information is acquired as information about vehicles: number of passing vehicles, average speed, time occupancy rate, and space occupancy rate. Furthermore, the following information is acquired as information about pedestrians: number of passing vehicles, number of stagnant vehicles (number of waiting vehicles), average speed, time occupancy rate, and space occupancy rate. Furthermore, the following information is acquired as information about bicycles: number of passing vehicles, number of stagnant vehicles (number of waiting vehicles), average speed, time occupancy rate, and space occupancy rate.

[0063] The example shown in Figure 7 is for the first monitoring area (straight-through left-turn lane area) and includes information about vehicle traffic conditions. The spatial occupancy rate is the percentage of space occupied by moving objects (vehicles, pedestrians, bicycles) within the observation area at the time of observation. The time occupancy rate is the percentage of time that a moving object is present at the observation point within the observation time.

[0064] Next, a description will be given of a schematic configuration of the signal control server 2. Fig. 8 is a block diagram showing a schematic configuration of the signal control server 2.

[0065] The signal control server 2 includes a communication unit 21, a storage unit 22, and a processor 23.

[0066] The communication unit 21 communicates with the camera 1 and the traffic light controller 31 .

[0067] The storage unit 22 stores programs executed by the processor 23 and the like.

[0068] The processor 23 performs various processes by executing programs stored in the storage unit 22. In this embodiment, the processor 23 performs image analysis processing, signal control processing, and the like.

[0069] In the image analysis process, the processor 23 extracts an image of the monitoring area from the camera image acquired by the camera 1, and acquires traffic condition information (real-time traffic flow measurement information) for each monitoring area based on the area image.

[0070] Specifically, for the first monitoring area (straight-going left-turn lane area), the tail position of the queue of vehicles in the straight-going left-turn lane, i.e., the length of the queue (distance to the tail vehicle), the number of vehicles going straight, and their average speed are acquired. For the second monitoring area (right-turn lane area), the tail position of the queue of vehicles in the right-turn lane, i.e., the length of the queue (distance to the tail vehicle), the number of vehicles turning right, and their average speed are acquired. For the third monitoring area (left-turn waiting area), the number of vehicles turning left and their average speed are acquired. For the fourth monitoring area (pedestrian crossing area), the number of pedestrians passing and their average speed, and the number of bicycles passing and their average speed are acquired. For the fifth monitoring area (crosswalk waiting area), the number of pedestrians and the number of bicycles remaining are acquired.

[0071] In the signal control process, the processor 23 generates signal control information based on the traffic condition information for each monitoring area acquired in the image analysis process. In this embodiment, the processor 23 identifies which of the following problems should be improved with the highest priority (priority problem): problems related to right-turning vehicles (see FIG. 2), problems related to straight-moving vehicles (see FIG. 3), and problems related to left-turning vehicles and crossing pedestrians (see FIG. 4), and generates signal control information to individually improve the identified priority problems.

[0072] Here, when generating signal control information from traffic condition information, a signal control model constructed by learning using past control results may be used. In this case, a signal control model is constructed that derives optimal signal control information that smooths the passage of vehicles and pedestrians from current signal control information by learning using traffic condition information at each time point in the past, signal control information used at that time, and control results at that time. Furthermore, a signal control model may be constructed for each time period.

[0073] Specifically, traffic conditions based on past traffic condition information are categorized into multiple patterns, a signal control model is constructed for each pattern, and the signal control information set for the pattern corresponding to the current traffic conditions is obtained as the optimal signal control information using this signal control model.In addition, a signal control model (machine learning model) is constructed using machine learning such as deep learning, and the signal control model is used to derive the optimal signal control information from the current traffic condition information.

[0074] In this embodiment, the signal control server 2 performs image analysis processing and signal control processing, but it is also possible to configure the image analysis server that performs image analysis processing to be provided separately from the signal control server 2 that performs signal control processing.

[0075] In addition, in this embodiment, the signal control server 2 performs a process of extracting images of each monitoring area from the captured image, but the camera 1 may perform a process of extracting images of each monitoring area from the captured image, and the images of each monitoring area may be transmitted to the signal control server 2.

[0076] In addition, in the signal control server 2, traffic condition information obtained by image analysis processing is provided to the signal control processing. At this time, if the traffic condition information is provided to the signal control processing as virtual detector information, there is no need to significantly modify the configuration of conventional signal control processing based on detector information.

[0077] Next, a description will be given of the procedure of the processing carried out by the signal control server 2. FIG.

[0078] First, the signal control server 2 acquires traffic condition information for each monitoring area by analyzing camera images at a predetermined control interval (e.g., 1 minute, 2.5 minutes, 5 minutes, 15 minutes, etc.) (ST101). Here, the traffic condition information includes the number of passing vehicles and their average speed, the number of passing pedestrians and the number of waiting pedestrians, etc. (See FIG. 7).

[0079] Next, the signal control server 2 calculates the balance (stability) of traffic flow at the intersection (ST102), by calculating the balance of traffic flow for each lane and crosswalk.

[0080] Next, the signal control server 2 determines whether the traffic flow at the intersection is stable based on the balance of the traffic flow (ST103). If the traffic flow is stable (Yes in ST103), the process returns to ST101.

[0081] On the other hand, if the traffic flow is unstable (No in ST103), the signal control server 2 then collects required information and identifies a priority issue (an issue that should be improved with priority) (ST104). Specifically, it is determined whether the priority issue is an issue related to right-turning vehicles, an issue related to straight-moving vehicles, or an issue related to left-turning vehicles and crossing pedestrians.

[0082] Here, if the priority issue is an obstacle related to a vehicle turning right ("turn right" in ST105), the process proceeds to the right turn improvement process (ST106) (see FIG. 10). If the priority issue is an obstacle related to a vehicle going straight ("go straight" in ST105), the process proceeds to the straight go improvement process (ST107) (see FIG. 11). If the priority issue is an obstacle related to a vehicle turning left and a crossing pedestrian ("turn left crossing" in ST105), the process proceeds to the left turn crossing improvement process (ST108) (see FIG. 12).

[0083] In the example shown in FIG. 9, the process is executed at a predetermined control period, but the process may be executed at the timing when one cycle of signal control ends.

[0084] Furthermore, in this embodiment, priority issues (issues that should be improved with priority) are identified, and control is implemented to individually improve the priority issues. However, without identifying priority issues, control may be implemented to comprehensively improve issues that should be improved, i.e., issues related to right-turning vehicles, issues related to straight-moving vehicles, and issues related to left-turning vehicles and crossing pedestrians. Furthermore, control to individually improve priority issues and control to comprehensively improve multiple issues may be implemented in combination. In this case, a determination to identify a priority issue and an evaluation of the overall traffic situation may be performed appropriately.

[0085] Next, a description will be given of the procedure of the right-turn improvement processing (ST106 in FIG. 9) performed by the signal control server 2. FIG. 10 is a flow diagram showing the procedure of the right-turn improvement processing.

[0086] When the priority issue (issue that should be improved with priority) is an obstacle to right-turning vehicles (such as congestion in the right-turn lane), the signal control server 2 executes a right-turn improvement process.

[0087] In the right turn improvement process, the signal control server 2 first determines which of the first to fifth observation positions the end of the jammed vehicle queue is at (ST201).

[0088] Here, if the end of the jammed vehicle line is the first observation position ("first observation position" in ST201), the degree of disruption is low, so the signal control server 2 generates standard signal control information (ST202).

[0089] Furthermore, if the end of the jammed vehicle queue is the second observation position ("second observation position" in ST201), the degree of disruption is low, so the signal control server 2 generates standard signal control information (ST203).

[0090] On the other hand, if the end of the jammed vehicle queue is at the third observation position ("third observation position" in ST201), the signal control server 2 generates signal control information (ST204) to shorten the vehicle green time (the signal time during which straight-moving vehicles are permitted to proceed) and extend the right-turn green time (the signal time during which only right-turning vehicles are permitted to proceed) to a relatively short time. Specifically, the right-turn green time is set to the standard time plus half of the extendable time.

[0091] Furthermore, if the end of the jammed vehicle queue is the fourth observation position ("fourth observation position" in ST201), the signal control server 2 generates signal control information to shorten the vehicle green time and extend the right-turn green time relatively long (ST205). Specifically, the right-turn green time is set to the standard time plus 2 / 3 of the extendable time.

[0092] Furthermore, if the end of the queue of vehicles in the traffic jam is the fifth observation position ("fifth observation position" in ST201), the signal control server 2 generates signal control information to shorten the vehicle green time and extend the right-turn green time to the maximum (ST206). Specifically, the right-turn green time is set to the standard time plus the extendable time.

[0093] In the right-turn improvement process shown in FIG. 10, a determination is made as to the end of the jammed line of vehicles, i.e., whether the end of the jammed line of vehicles is at any of the first to fifth observation positions. However, in the process (ST104) for identifying the priority issue shown in FIG. 9, a determination is made as to the end of the jammed line of vehicles, and if the end of the jammed line of vehicles is at any of the third, fourth, or fifth observation positions, it may be determined that the priority issue is an obstacle to right-turning vehicles, since there is significant congestion in the right-turn lane.

[0094] In addition, the extension time to be added to the standard time may be determined by categorizing the traffic situation into multiple patterns based on past traffic condition information (number of right-turning vehicles, average speed, etc.), setting the optimal extension time for each pattern, and adopting the extension time set for the pattern that corresponds to the current traffic situation.

[0095] Next, a description will be given of the procedure of the straight driving improvement processing (ST107 in FIG. 9) performed by the signal control server 2. FIG. 11 is a flowchart showing the procedure of the straight driving improvement processing.

[0096] When the priority issue (issue that should be improved with priority) is an obstacle to straight-moving vehicles (such as a backlog of straight-moving vehicles), the signal control server 2 executes straight-moving improvement processing.

[0097] In the straight driving improvement process, the signal control server 2 first determines whether the straight driving vehicle speed (average speed of straight driving vehicles) is equal to or less than a predetermined threshold value (for example, 20 km / h) (ST301).

[0098] Here, if the straight-moving vehicle speed is equal to or less than the threshold value (Yes in ST301), the signal control server 2 generates signal control information (ST302) to shorten the right-turn green arrow time (the signal time during which only right-turning vehicles are permitted to proceed) and to maximize the vehicle green time (the signal time during which straight-moving vehicles are permitted to proceed). This allows smoother passage for straight-moving vehicles and reduces the accumulation of straight-moving vehicles.

[0099] On the other hand, if the straight-moving vehicle speed exceeds the threshold value (No in ST301), the degree of interference is low, so the signal control server 2 generates standard signal control information (ST303).

[0100] In the straight-line driving improvement process shown in FIG. 11, a determination is made regarding the straight-line vehicle speed, i.e., whether the straight-line vehicle speed is equal to or less than a threshold value. However, in the process (ST104) for identifying a priority issue shown in FIG. 9, a determination is made regarding the straight-line vehicle speed. If the straight-line vehicle speed is equal to or less than a threshold value, it may be determined that the priority issue is an obstacle to straight-line vehicles, since the passage of straight-line vehicles is not smooth.

[0101] Next, a description will be given of the procedure of the left-turn crossing improvement process (ST108 in FIG. 9) performed by the signal control server 2. FIG. 12 is a flow diagram showing the procedure of the left-turn crossing improvement process.

[0102] When the priority issue (issue that should be improved with priority) is an issue related to left-turning vehicles and crossing pedestrians, the signal control server 2 executes left-turn crossing improvement processing.

[0103] In the left-turn crossing improvement process, the signal control server 2 first determines which of the first to fourth events the current traffic situation corresponds to based on the number of pedestrians crossing (the number of pedestrians passing by) and the left-turning vehicle speed (the average speed of left-turning vehicles) (ST401). Here, the first event is when there are many pedestrians crossing and the left-turning vehicle speed is high. The second event is when there are few pedestrians crossing and the left-turning vehicle speed is high. The third event is when there are many pedestrians crossing and the left-turning vehicle speed is low. The fourth event is when there are few pedestrians crossing and the left-turning vehicle speed is low.

[0104] Here, if the first event occurs, that is, if there are a large number of pedestrians crossing the road and the speed of left-turning vehicles is high ("first event" in ST401), the signal control server 2 generates signal control information to extend the pedestrian green time (the time the signal allows pedestrians to cross) (ST402). This gives priority to pedestrians crossing the road, and reduces the accumulation of pedestrians in the crossing waiting area.

[0105] Furthermore, if the second event applies, that is, if the number of pedestrians crossing the road is small and the speed of the left-turning vehicle is high ("second event" in ST401), the degree of disruption is low, so the signal control server 2 generates standard signal control information (ST403).

[0106] Furthermore, if the third event occurs, that is, if there are a large number of pedestrians crossing the road and the left-turning vehicle speed is low ("third event" in ST401), the signal control server 2 generates signal control information to extend the vehicle green time (the signal time during which left-turning vehicles are permitted to pass) and the pedestrian green time (ST404). By extending the vehicle green time, the reduction in the speed of left-turning vehicles is alleviated, facilitating the smooth passage of left-turning vehicles. Furthermore, by extending the pedestrian green time, the accumulation of pedestrians in the crossing waiting area is alleviated.

[0107] In addition, in the case of the third event, that is, when there are a large number of pedestrians crossing the road and the speed of left-turning vehicles is low, if the number of pedestrians crossing is extremely large, the pedestrians crossing the road are given priority over left-turning vehicles, and signal control information may be generated to extend the green time for pedestrians without extending the green time for vehicles.

[0108] Furthermore, if the fourth event occurs, that is, if the number of pedestrians crossing the road is small and the speed of left-turning vehicles is low ("fourth event" in ST401), the signal control server 2 generates signal control information to extend the green light time for vehicles (ST405). This gives priority to left-turning vehicles, facilitating their passage.

[0109] In the example shown in FIG. 12, traffic light control is performed by dividing the cases into those based on the traffic conditions of left-turning vehicles, specifically the number of pedestrians crossing the road, and those of pedestrians crossing the road, specifically the average speed of left-turning vehicles. However, traffic light control may also be performed by dividing the cases into those based on the waiting conditions of pedestrians waiting to cross the road, specifically the number of people waiting, in addition to those based on the traffic conditions of left-turning vehicles and those of pedestrians crossing the road.

[0110] Next, an outline of the signal control performed by the signal control server 2 will be described. Fig. 13 is an explanatory diagram showing a signal stage table showing the status of signal control.

[0111] Figure 13(A) shows a basic signal configuration at an intersection where two roads intersect. For example, extending the green light time for vehicles (1G, 2G) can encourage vehicle traffic and reduce congestion. Also, extending the green light time for pedestrians (1PG, 2PG) can encourage pedestrians to cross the street and reduce congestion of pedestrians waiting to cross.

[0112] Furthermore, for example, as shown in Figure 13(B), dividing the first phase 1φ into one that allows pedestrians to cross and one that does not allow pedestrians to cross can facilitate the passage of left-turning vehicles. Also, for example, as shown in Figure 13(C), adding a phase that allows only right-turning vehicles to pass can facilitate the passage of right-turning vehicles. Also, for example, as shown in Figure 13(D), adding a phase that only allows pedestrians to cross can encourage pedestrians to cross and reduce the accumulation of pedestrians waiting to cross.

[0113] As described above, the embodiments have been described as examples of the technology disclosed in this application. However, the technology in this disclosure is not limited to these, and can be applied to embodiments in which modifications, substitutions, additions, omissions, etc. are made. Furthermore, it is also possible to combine the components described in the above embodiments to create new embodiments. [Industrial Applicability]

[0114] The traffic signal control device of the present invention has the effect of being able to sufficiently increase the accuracy of control to improve congestion in the right-turn lane when congestion occurs in the right-turn lane, and also being able to sufficiently increase the accuracy of control to improve congestion of pedestrians waiting to cross or left-turning vehicles when congestion occurs, and is useful as a traffic signal control device that controls traffic signals installed at road intersections. [Explanation of symbols]

[0115] 1 camera 2. Signal control server (traffic signal control device) 3 traffic lights 21 Communications Department 22 Memory section 23 processors 31 Signal controller 32 Vehicle signal lights 33 Pedestrian signal lights

Claims

1. A traffic signal control device having a processor that executes a process of generating signal control information for controlling a traffic signal installed at an intersection based on traffic condition information around the intersection, The processor: Obtain the detection results of a sensor that detects objects on the road, acquiring traffic condition information by analyzing the detection results of the sensors, and generating signal control information based on the traffic condition information; In generating the signal control information, For each of a plurality of observation positions set within the monitoring area, the position of the end of the queue of vehicles in the right-turn lane that is jammed is acquired as the traffic condition information based on the occupancy status of objects on the road surface within a detection frame corresponding to each observation position, and the signal time for allowing right-turning vehicles to proceed is adjusted in stages depending on which observation position the end of the queue is at; A traffic signal control device that acquires the average speed of left-turning vehicles and the number of crossing pedestrians as traffic condition information based on the traffic conditions of left-turning vehicles and crossing pedestrians, and adjusts at least one of the signal time for allowing left-turning vehicles to proceed and the signal time for allowing pedestrians to cross depending on the combination of the average speed of left-turning vehicles and the number of crossing pedestrians.

2. The processor:

2. The traffic signal control device according to claim 1, characterized in that information regarding the speed of straight-moving vehicles is obtained based on the detection results of the sensor regarding the monitoring area targeting the straight-moving lane, and the signal time for allowing straight-moving vehicles to pass is adjusted based on that information.

3. The processor: The traffic signal control device according to claim 1, characterized in that it generates signal control information that facilitates the passage of vehicles and pedestrians from current traffic condition information using a signal control model constructed by learning using traffic condition information at each point in time in the past, signal control information used at that point in time, and control results at that point in time.

Images