Intersection environment perception and intelligent intersection system

Through the environmental perception and edge computing technology of multiple roadside smart base stations, combined with high-precision maps and traffic data, the refined management and information closed loop of urban road intersections is realized, which solves the problem of frequent traffic accidents and improves traffic safety Sex and efficiency.

CN119028119APending Publication Date: 2024-11-26VANJEE TECHNOLOGY CO LTD
View PDF 0 Cites 6 Cited by

Patent Information

Application Number
CN202310607977.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-26
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Traffic accidents occur frequently at urban road intersections, and it is difficult for the existing technology to achieve effective traffic management and information interaction, resulting in limited traffic capacity and safety.

Method used

Multiple roadside smart base stations are used for environmental perception, registration and synchronization is carried out through high-precision maps and GNSS signals, lidar and image data are obtained for fusion calculation, and environmental perception and smart intersection system at intersections are realized in combination with edge computing to provide real-time Traffic information and early warning services.

Benefits of technology

It improves traffic safety and traffic efficiency at intersections, reduces the incidence of traffic accidents, and realizes refined management and closed information loop of intersections.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure FT_1
    Figure FT_1
  • Figure FT_2
    Figure FT_2
  • Figure FT_3
    Figure FT_3
Patent Text Reader

Abstract

The invention relates to an intersection environment sensing method. The method comprises the following steps: acquiring sensing data of a plurality of roadside smart base stations; wherein the roadside smart base station is arranged at a preset position of an intersection; all the roadside smart base stations carry out registration based on a high-precision map, and all the roadside smart base stations carry out time synchronization based on GNSS signals; performing spatial synchronization on sensing data of each side base station by using the registration parameters; and calculating the sensing data after space synchronization to obtain an environment sensing result of the intersection. According to the method provided by the invention, refined global perception of the intersection area can be realized, and reasonable management and control of the intersection are facilitated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent transportation technology, and in particular to an environment perception and intelligent intersection system, computer equipment and storage medium for an intersection. Background Art

[0002] Urban road construction With the continuous progress and development of cities, urban roads are becoming more and more dense, and there are more and more road network nodes. Intersections have become an important part of the urban road smart intersection system. According to statistics, nearly 200,000 traffic accidents have occurred nationwide each year in recent years, of which about 70% occurred at intersections. Urban road intersections are at the intersection of pedestrians and vehicles. Due to factors such as conflicting vehicle driving routes and obstructed driver's vision when turning, traffic accidents at intersections are frequent. Therefore, the traffic capacity and traffic safety of urban road intersections have become a prominent problem affecting the traffic of urban road networks. How to reduce the incidence of traffic accidents at urban road intersections, improve the traffic capacity of urban road intersections, and enhance the sense of security of traffic participants has long been a problem that has plagued urban traffic management.

[0003] Urban traffic management uses comprehensive statistical knowledge, combines field investigation with simulation analysis, and continuously optimizes the traffic organization plan for urban road intersections to maximize the role of intersections. However, due to the conflict between vehicle and pedestrian rights of way, traffic chaos, traffic congestion, and frequent accidents at intersections are common. In response to the above problems, smart urban road intersections are built to digitally reconstruct people, vehicles, roads, and traffic environments, realize refined management of intersections, reduce traffic safety accidents at intersections, and alleviate traffic congestion pressure at intersections.

[0004] Currently, information such as intersection videos, traffic lights, electronic police, traffic collection, and event detection are all transmitted to the background center for processing. Traffic facilities at intersections cannot be horizontally interconnected or exchange information. Signal control and perception information push cannot form a closed loop at the intersection. At the same time, affected by the network, computing, storage, and latency, the refined management of intersections is restricted and affected to a certain extent. Summary of the invention

[0005] Based on this, it is necessary to provide intersection environmental perception, smart intersection systems, computer equipment and storage media to address the above technical issues.

[0006] An environment perception method for an intersection, comprising: Acquire sensing data of multiple roadside smart base stations; wherein the roadside smart base stations are deployed at preset positions of intersections; each of the roadside smart base stations is aligned based on a high-precision map, and each of the roadside smart base stations is time synchronized based on a GNSS signal; Use registration parameters to spatially synchronize the sensing data of each roadside base station; The spatially synchronized perception data is calculated to obtain the environmental perception result of the intersection.

[0007] As an optional implementation, the intersection is one of an intersection where a main road intersects with a main road, an intersection where a main road intersects with a secondary road, an intersection where a secondary road intersects with a secondary road, and an X-shaped intersection, and the roadside smart base station is arranged diagonally.

[0008] As an optional implementation, when the intersection is an X-shaped intersection, the roadside intelligent base station is deployed at the diagonal position where two roads are relatively closer to each other at the X-shaped intersection.

[0009] As an optional implementation, the intersection is also equipped with a roadside sound and light warning device for outputting guidance information, and the roadside sound and light warning device is set based on the crosswalk on the road.

[0010] As an optional implementation manner, the acquiring of the sensing data of multiple roadside intelligent base stations includes: Acquire lidar data and image data; Performing spatiotemporal synchronization on the laser radar data and the image data to obtain data to be fused; The data to be fused is fused and calculated to obtain the perception data of the roadside intelligent base station.

[0011] As an optional implementation manner, the acquiring of the sensing data of multiple roadside intelligent base stations includes: Acquire millimeter wave radar data and image data; Performing spatiotemporal synchronization on the millimeter wave radar data and the image data to obtain data to be fused; The data to be fused is fused and calculated to obtain the perception data of the roadside intelligent base station.

[0012] As an optional implementation, the perception data includes attributes, speed, heading angle and identity of traffic participants at the intersection.

[0013] A smart intersection system, comprising: Multiple roadside smart base stations are used to obtain sensing data; wherein the roadside smart base stations are deployed at preset positions at intersections; each of the roadside smart base stations is aligned based on a high-precision map, and each of the roadside smart base stations is time synchronized based on a GNSS signal; The edge computing unit is used to spatially synchronize the perception data of each roadside base station using the alignment parameters; and calculate the spatially synchronized perception data to obtain the environmental perception result of the intersection.

[0014] In a third aspect, an embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method of the first aspect when executing the computer program.

[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method of the first aspect described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of widening a flat intersection to provide a left-turn lane in one embodiment; Figure 2 A schematic diagram of widening a flat intersection to provide a right-turn lane in one embodiment; Figure 3 A schematic diagram of a smart intersection system architecture in one embodiment; Figure 4 A schematic diagram of a laser radar confirming a suspected small target in one embodiment; Figure 5 A schematic diagram of a city traffic road network intersection range in one embodiment; Figure 6 A schematic diagram of a deployment method of a single base station at a smart intersection in one embodiment; Figure 7 A schematic diagram of dual base station deployment at a smart intersection in one embodiment; Figure 8 A schematic diagram of multiple base stations deployed at a smart intersection in one embodiment; Fig. 9 A schematic diagram of the layout of intersection equipment where a trunk road intersects a trunk road in one embodiment; Fig.10 A schematic diagram of the layout of intersection equipment where a trunk road intersects a trunk road in another embodiment; Fig.11 A schematic diagram of equipment layout at an intersection where a main road and a secondary road intersect in one embodiment; Fig.12 A schematic diagram of equipment layout at an intersection where a main road and a secondary road intersect in another embodiment; Fig.13 A schematic diagram of equipment layout at an intersection where a secondary trunk road intersects a secondary trunk road in one embodiment; Fig.14 A schematic diagram of equipment layout of an X-shaped intersection schematic diagram in one embodiment; Fig.15 A schematic diagram of equipment layout of a T-shaped intersection schematic diagram in one embodiment; Fig.16A schematic diagram of a network architecture of a smart intersection system in one embodiment; Fig.17 A schematic diagram of signal control induction for a special vehicle in an embodiment. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0018] Smart intersection: traffic information that comprehensively represents intersection traffic. Smart intersections use laser radar, cameras and other sensing methods, integrate sensor technology, high-precision map technology, AI artificial intelligence technology, and edge computing technology to perceive intersection traffic information in all time and space, multiple dimensions, and all elements, conduct information interaction between people, vehicles, and roads, realize refined management of intersections, ensure traffic safety, and improve traffic efficiency and management level.

[0019] Urban road intersections are divided into flat intersections and grade-separated intersections. The scheme of this application mainly involves flat intersections. Flat intersections are divided into Class A (signal-controlled intersections), Class B (intersections without signal control), and Class C (roundabout intersections) according to the traffic organization method. Among them, Class A is further divided into Class A1 (intersections controlled by traffic signals, with widened entrance roads), Class A2 (intersections controlled by traffic signals, with non-widened entrance roads); Class B is further divided into Class B1 (intersections where the center of the main road is isolated and closed, and only right turns are allowed on branch roads, referred to as right-turn intersections), Class B2 (intersections controlled by slow down and give way or stop and give way signs, referred to as give-way intersections), and Class B3 (intersections with no control at all).

[0020] According to the type of intersecting roads, planar intersections are divided into six types: main road-main road (mostly Class A1), main road-secondary road (mostly Class A1), main road-branch road (mostly Class B1), secondary road-secondary road (mostly Class A1), secondary road-branch road (mostly Class B2), and branch road-branch road (mostly Class B2 or Class B3).

[0021] According to the geometric shape, the plane intersection can be divided into cross, T, Y, X, multi-fork, staggered and roundabout intersections. In this plan, the cross, X and T intersections are selected for equipment deployment. For other types of intersections, you can refer to the diagram in the plan. Special intersections are designed according to the actual situation. Among them, the main road: 30m~40m, the design speed of the main road is 40-60km / h. Secondary road: 20m~24m, the design speed of the secondary road is 30-50km / h. Branch road: 14m~18m, the design speed of the branch road is 20-40km / h.

[0022] At grade intersections, the traffic capacity can be improved by adopting methods and measures such as dedicated left, straight and right turn lanes for motor vehicles, dedicated right turn lanes for non-motor vehicles, widening of entrance lanes, offsetting the center line of entrance lanes, compressing the width of the central dividing strip of entrance lanes, ahead waiting for motor vehicles to turn left, secondary crossing of pedestrians, traffic signal control phase plans, traffic signs and markings, and traffic separation and diversion facilities.

[0023] The width of an entrance lane at a flat intersection should be 3.25m. In difficult situations, the minimum width can be 3.0m. When the land for reconstruction of the intersection is limited, the minimum width of an entrance lane can be 2.80m. The right-turn lane on the right side of the corner guide island should be widened according to the design speed and turning radius.

[0024] The entrance road length consists of the length of the widening gradient section (Lt) and the widening section (Ld). The length of the gradient section (Lt) is calculated when a vehicle moves across a lane at 70% of the design speed of the road section for 3 seconds. The minimum length of the gradient section should not be less than: 20m for branch roads, 25m for secondary roads, and 30m~35m for main roads. The minimum length of the widening section should ensure that left-turning or right-turning vehicles are not affected by the length of the adjacent waiting vehicles.

[0025] Figure 1 The illustration shows a widening of the intersection to set up a left-turn lane. Figure 1 In the formula, Lt: the length of the gradient section where the lane is changed (m), Ls: the length of the adjacent waiting vehicle queue (m); Ld: the length of the widening section and the deceleration lane (m); Lz: the minimum length of the dedicated left-turn lane (m). In the right-turn lane, the length of the adjacent waiting vehicle queue (Ls) can be determined by the following formula: Ls=9N Where: N is the number of vehicles queuing for left or right turns during each signal cycle within 15 minutes of the peak hour.

[0026] When two special lanes for turning are required, the length of the widening section can be 60% of the length of one special lane. When there is no traffic volume data, the minimum length of the widening section should not be less than: 30m~40m for branch roads, 50m~70m for secondary roads, and 70m~90m for main roads. The lower limit is taken when it intersects with branch roads, and the upper limit is taken when it intersects with main roads.

[0027] Figure 2 The illustration shows a widening of the intersection to set up a dedicated right-turn lane. Figure 2Lt: length of gradient section (m); Ld: length of widening section, not less than the length of adjacent lane vehicle queue (m); Lz: minimum length of dedicated left-turn lane (m); la: distance required for vehicle acceleration (m); ly: length of widened right-turn dedicated lane (m); Ly': length of widened acceleration lane (m). Among them, the width of each lane of the exit road should not be less than the lane width of the road section, preferably 3.5m. The width of each lane of the exit road of the rebuilt intersection with limited conditions should not be less than 3.25m.

[0028] The length of the exit road consists of the widening section and the widening gradient section. The minimum length of the widening section should not be less than 30m~60m. The upper limit is used for trunk roads with heavy traffic, and the lower limit can be used for other roads. The minimum length of the gradient section should not be less than 20m.

[0029] In one embodiment of the present application, a smart intersection system architecture for realizing a smart intersection is provided. The smart intersection system architecture involves five layers: device layer, communication layer, data layer, platform layer, and application layer. For details, see Figure 3 .

[0030] Among them, the equipment layer includes perception equipment, traffic control equipment and warning equipment. The perception equipment is mainly composed of roadside smart base stations (also known as roadside fusion perception smart intersection system or roadside base station or smart base station), edge computing units, etc., through reasonable deployment to achieve full coverage detection of intersections, and real-time perception of traffic participants' types, positions, speeds, sizes, directions, predicted trajectories and other information; Among them, the roadside smart base station can be a perception smart intersection system composed of multiple sensors, and the perception smart intersection system can run relevant algorithms to process data collected by multiple sensors to achieve environmental perception. The algorithm includes but is not limited to the target detection algorithm corresponding to the sensor data, the multi-sensor fusion perception algorithm, etc. The sensor on the roadside smart base station can be one or more of laser radar, millimeter wave radar, and high-definition camera. The roadside smart base station can provide large-scale, multi-behavior, and multi-target perception capabilities, and realize the recognition of all road traffic elements based on the fusion technology of laser radar and video. It can achieve perception within a radius of at least 80m, and has the perception capabilities of traffic flow status, traffic events, and traffic participants. When the roadside smart base station is a fusion perception of millimeter wave radar and video, its coverage detection distance will be farther. When laying out actual road scenes, for straight areas, millimeter-wave radar and video can be used in combination for road perception, while for complex road scenes (such as intersections), lidar and video can be used in combination to ensure the detection rate, detection success rate or tracking success rate of traffic participants in complex scenes.

[0031] Among them, the communication layer realizes the real-time transmission of various traffic information. The communication layer can communicate through sensor terminals mainly based on V2X, or through wireless or wired bearer networks such as cellular mobile networks (3G / 4G / 5G) or optical fiber communications. Of course, multiple types of communication networks can also be selected for combination. Optionally, the smart intersection system architecture provided by this application can make full use of edge computing technology to analyze and process perception information, and transmit information to the cloud with extremely low latency through optical fiber or 4G / 5G communication, realizing "people-vehicle-road-cloud" collaborative interaction.

[0032] Among them, the role of the data layer is to build a digital middle platform for smart intersections. Through multi-source data fusion, three-dimensional high-precision maps, big data, cloud computing, the Internet of Things (IoT), artificial intelligence (AI) and other advanced digital technologies, a digital intersection (intersection digitization) that is "twin" of the physical intersection is simultaneously formed.

[0033] The digital middle platform for smart intersections is a complex smart intersection system that integrates multiple units such as unified data standards, intersection information models, intersection perception data (traffic information, vehicle information and signal light data), common support platforms, and digital twin applications. The spatiotemporal positioning of traffic participants is the most important link, which enables the simultaneous construction of physical models within the intersection detection range as digital twins. The digital middle platform is a digital model for the construction of smart intersections, and it is also a supporting platform for the full-process smart application of intersections. Relying on the data middle platform, it is highly integrated with information technologies such as holographic perception, simulation, and deep learning to carry out all-round and multi-dimensional construction of all-road intersection applications, becoming an important driving force for the refined management of digital intersections.

[0034] The platform layer mainly includes the smart intersection system for intersection situation awareness, the smart intersection system for operation status assessment and analysis, the smart intersection system for traffic information release, the smart intersection system for traffic signal optimization and collaboration, and the smart intersection system for vehicle-road collaboration.

[0035] Situational awareness: including holographic perception of intersections, digital reconstruction of identities, recording of violations, restoration of traffic accidents and data statistics, to achieve business functions such as situational awareness, statistics and display of smart intersection operations.

[0036] Research and analysis: including operation status analysis, traffic safety analysis and analysis, and traffic status evolution law analysis. Based on the front-end multi-source fusion perception detection data and combined with business needs, the operation status, traffic safety and traffic flow evolution law of the intersection are analyzed and judged to provide data support for intersection traffic control.

[0037] Information release: including traffic incident warnings, traffic data inquiries, traffic status prompts and violation management. The smart intersection system releases event warnings, violation reminders, traffic status prompts, violation management and other information to traffic participants within the detection range of the smart intersection through roadside information release and sound and light warning facilities, so as to improve the traffic order at the intersection and enhance the traffic efficiency.

[0038] Signal collaborative optimization: includes traffic signal detection and evaluation and traffic signal collaborative optimization. Based on front-end multi-source fusion perception technology, it accurately obtains multi-dimensional signal evaluation indicators, conducts comprehensive diagnosis and monitoring of the current traffic signal operation status at intersections, and provides high-precision signal collaborative optimization parameters based on advanced technologies such as multi-source fusion perception, high-precision maps, and spatiotemporal positioning, thereby improving the efficiency of intersection signal optimization and the level of intelligent intersection signal timing.

[0039] Vehicle-road collaboration: including safety warning, perception supplement and abnormal status warning. The smart intersection system is based on the sharing of holographic stereoscopic perception data at the intersection. It supports the implementation of safety warning, beyond-visual-range / blind-spot perception compensation and abnormal status prompt applications for intelligent connected vehicles and autonomous driving vehicles in mixed traffic scenarios on urban roads.

[0040] Optionally, the roadside smart base station of the present application integrates a roadside laser radar (which can be a 360° omnidirectional laser radar), an AI camera, and an edge computing unit to form an integrated sensing device. The multi-source sensor fusion device integrates road information perception and data storage and computing functions to achieve accurate perception of traffic information, analyze and process the perception information using edge computing technology, and provide high-quality road information services. Optionally, the smart intersection in the present application adopts a roadside stereoscopic perception AI model technology based on multimodal cross-spectrum, which performs multivariate data fusion and feature extraction on the multimodal cross-spectrum perception data collected in real time. When the result-level data fusion of the point cloud data obtained by the laser radar and the video stream data obtained by the camera is performed, the traffic participant perception information is extracted from the video stream data and the laser point cloud data respectively, and then the spatial coordinate smart intersection system and time synchronization are realized, and then the target feature mapping, information matching association, data fusion correction, and finally the fused perception information is output. Based on the synchronization of spatiotemporal references, the precise positioning of multiple targets of traffic participants is achieved by means of perception calibration, noise reduction, target deduplication, algorithm fusion, etc., and the continuous tracking of the same target trajectory is achieved. On the basis of continuous tracking, a cross-time domain traffic safety risk warning algorithm and model based on swarm intelligence is established. Through the precise positioning and continuous tracking of multiple targets of traffic participants, the behavior information of traffic participants is obtained, and cross-time domain analysis of swarm intelligence is carried out to establish a traffic safety risk hazard warning algorithm, realize the detection of traffic accidents, illegal lane changes, speeding, slow driving, and wrong driving, analyze and judge the real-time safety risks of the road network, and improve the safety risk warning capability of road network nodes.

[0041] In one of the optional embodiments, after a traffic participant enters a scene area (which may be an intersection, or a scene area such as a tunnel or a bridge), a binding operation may be performed on the traffic participant to systematically associate the information of the traffic participant in various perception or communication systems. On this basis, the smart intersection system uses the deployed roadside smart base stations to track the targets in the scene area with identities, which is conducive to differentiated induction of traffic participants who are not very common at the intersection. In order to ensure the identity transmission of traffic participants between roadside smart base stations in these scenarios, the following method can be used to ensure it, including: when a traffic participant enters the overlapping area of ​​two roadside base stations, the position that the traffic participant is about to arrive is predicted based on the target tracking algorithm to obtain the predicted position, the predicted position is matched with the target detected by the next roadside base station, and the identity tag of the traffic participant is transmitted based on the matching result, and the identity tag may be a license plate number. Optionally, two adjacent smart base stations on the path of the traffic participant target can be used to match the tracking trajectory of the target in the overlapping area, and the identity tag is transmitted to ensure the accuracy of the identity transmission. Optionally, in order to ensure the accuracy of the matching, one or more information such as color, vehicle type, speed, heading angle, etc. may also be considered during the matching.

[0042] The traffic control equipment of the smart intersection system of the present application may include a traffic signal machine with intelligent networking function. According to the real-time traffic conditions at the intersection, the traffic signal is optimized and reasonably timed by the traffic signal controller with intelligent networking function to achieve traffic diversion. At the same time, priority access strategies are implemented for special vehicles (such as fire trucks, ambulances, etc.) and public transportation vehicles.

[0043] The warning equipment of the smart intersection system of the present application may include a variable information screen, sound and light alarm equipment, a voice announcer, etc., which can provide multi-mode warnings for pedestrians and non-motor vehicles at intersections through text, sound, and warning lights.

[0044] After one of the embodiments, the present application provides a method for improving traffic order at intersections. The method is developed from two aspects. On the one hand, centimeter-level high-precision maps and holographic perception smart base stations are used to collect the operating status of traffic participants in real time, combined with traffic signal access rights, to monitor illegal driving behaviors of motor vehicles, non-motor vehicles and pedestrians; on the other hand, the smart intersection system is based on precise fusion result data to establish a digital twin intersection with high unity in time and space, and to analyze in real time the conflict points of traffic flow at intersections, the common locations, types, and quantities of uncivilized phenomena, to support the precise control of traffic management departments, save police resources, and improve traffic order at intersections.

[0045] First, the roadside smart base station accurately senses the road traffic conditions and obtains centimeter-level location information, operating speed, driving trajectory, parking location and other information of traffic participants in all directions in real time.

[0046] Secondly, through the digital twin intersection, the three-dimensional visualization of the operation status of the intersection can be realized. At the same time, based on the preset AI analysis algorithm, the violations at the intersection can be monitored in real time. These violations may include frequent lane changes of motor vehicles, driving on the wrong road, non-motor vehicles driving in the opposite direction, running red lights, pedestrians running red lights, illegal entry, etc. Optionally, the smart intersection system can collect static traffic information at the intersection, including data such as signs and markings, traffic organization, channelization information, signal control strategies, and intersection environment information, and combine the roadside smart base station to perceive the information of all traffic participants at the intersection in real time. After unified alignment in time and space, the digital twin of the smart intersection is realized. Based on the preset AI analysis model, the intersection machine non-conflict points, traffic intersection points, and the common locations / numbers / types of violations are analyzed and sorted out, providing data support for the traffic management department to optimize the road organization form, refined management of intersections, and deployment of police forces at key locations.

[0047] With the support of the above-mentioned technologies, any detected violations or situations that require induction that have been analyzed can be reported, dissuaded and guided through communication with intelligent connected vehicles (which can be in the form of V2X, 5G, etc.) and / or roadside sound and light warning equipment to regulate traffic order.

[0048] In one of the optional embodiments, the present application provides a method for ensuring traffic safety at intersections. The method can be developed from two aspects. On the one hand, high-precision holographic sensing and low-latency information release equipment are used to ensure the safety of traffic participants. On the other hand, managers monitor and manage traffic events with high reliability, low latency, and three-dimensional visualization.

[0049] The information released can be: on the basis of realizing smart intersections (digitalization), based on high-precision maps, multi-source data fusion perception, spatiotemporal positioning, and digital twin technology, real-time monitoring of the safe driving behavior of traffic participants within the intersection range, and for unsafe driving behavior of traffic participants, the smart intersection system will issue early warnings and alerts through communication with vehicles and / or sound and light warning facilities on the roadside, thereby improving the safe traffic environment at the intersection.

[0050] The smart intersection system can report unsafe driving behaviors of motor vehicles, violations of pedestrians and non-motor vehicles to the traffic management department in real time, and realize three-dimensional visual motion trajectory presentation based on the digital middle platform, strengthen the control of illegal traffic behaviors of motor vehicles, non-motor vehicles and pedestrians, and improve the efficiency of intersection safety control of traffic management departments.

[0051] In one of the embodiments, the present application provides a method for improving the traffic efficiency at intersections, the method comprising: accurately sensing the road conditions through a roadside smart base station, obtaining real-time road traffic flow, vehicle speed, lane occupancy, headway, and vehicle queue length and other traffic flow information, facing traffic signal timing optimization and digital twin simulation scenarios, the roadside intelligent computing device provides IT service environment and computing power at the edge of the mobile network, and reports processed structured data at the same time. The traffic status data of key road nodes are aggregated from the edge nodes to the traffic center cloud for traffic big data analysis and data sharing, and the algorithm model is trained and upgraded at the same time, and the upgraded algorithm is pushed to the front-end roadside intelligent computing equipment for update and upgrade. The central cloud computing and edge computing work closely together to realize the digital upgrade of traffic management such as traffic light timing optimization and intersection digital twins based on the real-time traffic status perceived at the intersection, improve the traffic environment at the intersection, and improve the traffic efficiency at the intersection.

[0052] At the same time, the smart intersection system supports the transmission of intersection traffic information data to the central platform, and publishes information such as road conditions, events, congestion, etc. through the central traffic induction release system, providing timely and accurate traffic information services for traffic participants.

[0053] In one of the embodiments, the present application also provides a method for serving autonomous driving, the method comprising: based on sharing holographic stereoscopic perception data at intersections, solving the ultra-long line-of-sight and non-line-of-sight perception problems of autonomous driving vehicles in urban road scenes, and supporting the implementation of high-level autonomous driving in mixed traffic flow scenarios.

[0054] In one embodiment, the present application provides a target detection method for an intersection, which uses a laser radar to detect small targets at the intersection, ensuring accurate and efficient detection of traffic participants (traffic participants can be regarded as detection targets) in the intersection area. The method includes: When it is determined that there are suspected small targets at the intersection, the resolution of the laser radar is adjusted, and the area suspected of having small targets is scanned in a targeted manner. According to the point cloud of the targeted scan, the results of the small target detection are obtained. First, based on the high-precision map of the intersection area, the non-motor vehicle and pedestrian areas can be determined, and these areas can be detected to see if there are suspected small targets. In most cases, when there are targets in the non-motor vehicle lane and pedestrian area, the target that appears can be judged as a suspected small target. Of course, these areas sometimes change with the rule settings. For example, changes in traffic lights at crosswalks will cause changes in the type of targets passing through crosswalks. Small targets refer to pedestrians, non-motor vehicles and other targets on the road in traffic scenarios. These targets are relatively small compared to vehicles. When various targets confront each other in traffic scenarios, these targets are in a disadvantaged position. Sometimes there is a corresponding relationship between small targets and disadvantaged traffic participants.

[0055] Of course, in traffic scenarios, there may be pedestrians or non-motor vehicles appearing in motor vehicle lanes. In this case, it is possible to determine whether there are small targets among the detected targets based on the relative size of the detected targets (comparing adjacent targets, or comparing based on global detected targets). Optionally, the detection results of each area in the road can be divided and then compared to determine suspected small targets. Detection equipment such as bayonet cameras and lidars can be installed at preset positions at intersections. These devices collect information on targets that are about to enter the intersection area, and confirm whether there are suspected small targets among the targets entering the intersection area based on the collected information. These small targets are then scanned in a targeted manner to determine whether they are small targets. If they are confirmed to be small targets, they are tracked, and the parameters of the radar equipment deployed at the intersection are adjusted based on the tracking to enable targeted detection of small targets. Optionally, when the installed device is a lidar, a height-marking method (mechanical lidar) can be used to determine suspected small targets. For the principle, see . Figure 4 .

[0056] There is a geometric relationship between the detection light emitted by the laser radar and the height of the target, so the suspected small targets after the intersection can be screened out based on this. Of course, this screening method can also be combined with the areas delineated in the high-precision map to ensure the accuracy of the screening results. After determining that it is a small target, the trajectory of the small target can be predicted based on the tracking results, and the area to be reached can be obtained. The radar equipment corresponding to the area can be controlled and its resolution can be adjusted to perform targeted scanning on the area where the small target is about to appear, and the tracking results of the small target can be obtained. Two sets of mechanical laser radars can be arranged vertically at the bayonet, one of which is used to scan at a preset frequency, and the other is used to trigger when it is determined that there is a suspected small target, so as to scan the area of ​​the suspected small target in a targeted manner. Finally, the point cloud obtained by the two sets of laser radars is fused to obtain a point cloud with increased density, and whether the suspected small target is a small target is determined based on the point cloud with increased density. Optionally, when the laser radar is a MEMS laser radar, the focused scanning of suspected small targets can be achieved by controlling the deflection of the MEMS micromirror, specifically: the MEMS micromirror is arranged in the light beam receiving and transmitting direction of the light splitting component, and is triggered by the deflection control signal to reciprocate along the X-axis and reciprocate along the Y-axis to dynamically deflect the emission angle of the modulated pulse light beam emitted to the object to be measured and form a two-dimensional scanning field of view, and reflect the echo laser beam in the same direction back to the multiple detector pixels of the laser receiving component. Optionally, the laser control signal and the deflection control signal can be output respectively by the control circuits respectively connected to the laser emitting component, the laser receiving component and the MEMS micromirror, and the distance information of the object to be measured can be determined according to the electrical signal. The MEMS laser radar also includes: a laser transceiver component, the laser transceiver component includes a laser emitting component and a laser receiving component corresponding to the laser emitting component, the laser emitting component is triggered by a laser modulation signal to emit a modulated pulse beam, the laser receiving component is used to receive the echo laser beam and convert it into an electrical signal, wherein the laser receiving component includes a plurality of correspondingly distributed detector pixels; a spectroscopic component, used to change the relay path of the echo laser beam that has the same path as the modulated pulse beam in the detection space. When the laser radar is a FLASH laser radar, a turntable (deflection mechanism) needs to be configured for the FLASH laser radar, the FLASH laser radar performs a global scan according to the settings, and determines the key scanning area according to the comparison result of the target in the scanning result; adjust the field of view of the FLASH laser radar, and control the turntable at the same time so that the adjusted field of view of the FLASH laser radar just covers the area that needs to be scanned in a targeted manner, so as to achieve targeted detection of suspected small targets, and finally confirm whether it is a small target.When the laser radar is an OPA laser radar, the following scheme can be considered to realize the detection of traffic participants in the intersection, specifically including: scanning the preset scanning area with a first scanning resolution to obtain the first point cloud of the current cycle; performing target detection on the first point cloud of the current cycle, and if the target to be detected is detected, determining the fine scanning area and the observation scanning area based on the target to be detected; wherein the fine scanning area corresponds to the position of the target to be detected in the preset scanning area; the observation scanning area is the area of ​​the preset scanning area other than the fine scanning area; in the next scanning cycle, scanning the fine scanning area with a second scanning resolution, and scanning the observation scanning area with a third scanning resolution to obtain a second point cloud; wherein the second scanning resolution is greater than the first scanning resolution, and the third scanning resolution is less than or equal to the first scanning resolution. Optionally, the fine scanning area is an area corresponding to multiple positions of the multiple targets to be detected in the preset scanning area. Optionally, the method also includes: performing target detection on the second point cloud to obtain a fine detection result of the target to be detected. Optionally, if a new target appears in the detection result of the second point cloud, the new target is used as the target to be measured, and the fine scanning area and the observation scanning area are updated based on the target to be measured. Optionally, if no new target appears in the detection result of the second point cloud, the preset scanning area is scanned using the first scanning resolution in at least one cycle until the target to be measured is detected.

[0057] In one embodiment of the present application, another method for detecting small targets at intersections is provided, the method comprising: This method relies on high-precision map information for small target detection. First, the area of ​​interest for small target detection is defined. The road information of the high-precision map can be used to define the areas where small targets such as non-motor vehicles and pedestrians will appear or often appear as areas of interest. These areas of interest can be roadside straight lanes, intersection waiting areas, zebra crossing areas, and other areas.

[0058] When a target appears in the area, the corresponding detection process is started as follows: When the target enters the detection area, the point cloud data of the area is read, and the high-precision map is used to filter the roadside background in the point cloud data, such as trees, obstacles and other objects that block the background, to obtain the filtered regional point cloud to help detect small targets; When a target enters the detection area, the target detection network is used to detect the point cloud data of the area to obtain n consecutive frames of detection results; the target detection network can be a voxel-based 3D target detection network.

[0059] According to the association results of each target in the previous n-1 frames, the position of the target in the nth frame is predicted, and compared with the detection result of the nth frame to determine whether the detection result of the nth frame is correct; The judgment results are mainly divided into three situations: correct detection, false detection, and missed detection. If the result is a false detection, the detection target in the nth frame will be deleted; if the result is a missed detection, the predicted result will be used as the detection result of the current frame.

[0060] Optionally, in order to further improve the detection accuracy of small targets, the difference between the moving speed of small targets and the moving speed of vehicles on the road is considered. The small target moves slowly, and based on the detection position of the target in each frame of the first n-1 frames, the area where the target may appear in the next frame is roughly estimated, and the range is expanded as the search range of the target in the next frame, reducing the subsequent associated search range; Using the attention mechanism, the data of the detection frames of the previous n-1 frames are associated to obtain the status of the same target in the consecutive n-1 frames. On this basis, the position of each target in the n-1 frames is predicted according to the association results of the previous n-1 frames, and compared with the detection results of the n-1 frame to determine whether the detection result of the n-frame is correct. The judgment results are mainly divided into three situations: correct detection, false detection, and missed detection. If the result is a false detection, the detection target in the n-1 frame is deleted; if the result is a missed detection, the prediction result is used as the detection result of the current frame.

[0061] In addition, based on the small target detection method provided in the present application, the detection of small targets can be further optimized. For example, data enhancement can be considered for different scenarios; the detection results of laser and point cloud can be integrated; the output threshold of small target detection results can be lowered to obtain more small target detection frames and then track them; the tracking results can be post-processed, etc.

[0062] The small target detection method at the intersection realizes a high-precision tracking solution, so it can monitor the overall driving process of the small target and give a warning when the small target leaves the safe area or an abnormality occurs during driving. At the same time, the solution of the present application supports real-time statistics on the number and location of small targets in each area of ​​interest at the intersection, so as to manage the small targets in an orderly manner and reduce the occurrence of accidents.

[0063] This application scheme is mainly applicable to three types of plane intersections, namely, main road-main road, main road-secondary road, and secondary road-secondary road, that is, intersections that mainly adopt the A1 type of plane traffic organization method.

[0064] By installing and deploying roadside smart base station equipment at urban road intersections, full coverage detection of intersections can be achieved. This application provides corresponding installation and deployment plans according to the type of intersection and the installation of traffic lights. It should be noted that the deployment plan provided by this application sets the perception range of the roadside smart base station to within a radius of 80m, so the following intersection control area uses this as a reference standard. If the intersection is large and there are obvious obstructions, the number and location of equipment deployment need to be adjusted according to actual conditions.

[0065] Plane intersections are divided into straight, left turn and right turn. The construction of smart intersections needs to be designed as a whole based on the traffic volume of the intersection, the grade of the intersecting roads, the regional location of the intersection and the geographical conditions.

[0066] This application scheme is mainly applicable to three types of plane intersections with heavy traffic such as main road-main road, main road-secondary road, and secondary road-secondary road with large pedestrian and vehicle flow, that is, intersections that mainly adopt the A1 type of plane traffic organization method.

[0067] According to the design principles of Class A1 traffic intersections, the number of equipment deployed varies according to the size of the traffic intersection and the coverage of the equipment, including single base station deployment at the intersection, dual base station deployment at the intersection, and three base station deployment at the intersection. Figure 5 (Schematic diagram of the intersection range of urban traffic road network) is shown as follows: The intersection range in urban road traffic is defined as the intersection of the roads constituting the intersection and the area surrounded by the vertical lines where the import motor vehicle guide lane line is a solid line. The detection range is Figure 5 The area enclosed by the solid line.

[0068] Vehicles in different directions within the detection range of the intersection have completed lane changes, the traffic flow tends to be stable, and can cover non-motor vehicles and pedestrian traffic areas; basically, it can cover facilities and equipment related to the intersection. According to the design standards and the detection range of the smart base station, the relationship between the number of equipment deployed at intersections of different sizes is calculated as shown in the following table:

[0069] Where L is the distance from the center of the intersection to the farthest guide lane line in the four directions, and R is the detection radius of the roadside smart base station.

[0070] Based on experience, most secondary intersections are suitable for single base station deployment (see Figure 6 ), the main-secondary intersection is suitable for dual base station deployment ( Figure 7 ), main-main intersections are suitable for multiple base station deployment ( Figure 8 ).

[0071] In one embodiment, see Figure 9-Figure 15, common intersections can generally adopt the dual base station fusion solution of roadside smart base stations to meet the full coverage of the intersection. The details are as follows: Fig. 9 What is shown is a schematic diagram of equipment layout at an intersection where a main road intersects another main road. The solution is to diagonally deploy the roadside smart base station 100. The smart base station 100 can be deployed close to the crosswalk area, and the roadside sound, light and text warning equipment 200 is deployed at the corresponding intersection. Fig.10 Another schematic diagram of equipment layout at an intersection where a trunk road intersects a trunk road is shown, which also adopts diagonal layout of the roadside smart base station 100. In this solution, the roadside smart base station 100 is arranged in the middle area of ​​the intersection. The roadside sound, light and text warning equipment 200 is installed on both sides of the crosswalk.

[0072] Figure 11-Figure 12 Both of them show a schematic diagram of equipment layout at an intersection where a main road and a secondary road intersect. Both schemes use diagonal layout of roadside smart base stations 100. In this scheme, the roadside smart base stations 100 are arranged in the middle area of ​​the intersection. Roadside sound, light and text warning equipment 200 is installed on both sides of the crosswalk.

[0073] Fig.13 The diagram shows the equipment layout of a secondary trunk road-secondary trunk road intersection. Both schemes use diagonal road-side intelligent base stations 100. In this scheme, the road-side intelligent base station 100 is located in the middle area of ​​the intersection. Road-side sound and light text warning equipment (an information release device) 200 is installed on both sides of the crosswalk.

[0074] Fig.14 The diagram shows the equipment layout of an X-shaped intersection. The solution deploys a roadside smart base station 100 at the diagonal point where the two roads are closer to each other at the X-shaped intersection, and the roadside sound, light and text warning equipment 200 is installed on both sides of the crosswalk. The roadside smart base station can sense the roads in each direction at the X-shaped intersection, identify vehicles, pedestrians, and non-motor vehicles, and output accurate latitude and longitude coordinates, heading angles, speeds, etc., and can be used for scene applications based on the fused sensed information.

[0075] Fig.15 The diagram shows the equipment layout of a T-shaped intersection. In this solution, a roadside smart base station 100 is deployed on the roadside of the intersection of the T-shaped intersection, and a roadside sound, light and text warning device 200 is installed on the roadside of the crosswalk. The roadside smart base station can sense the roads in three directions at the T-shaped intersection, identify vehicles, pedestrians, and non-motor vehicles, and output accurate latitude and longitude coordinates, heading angles, speeds, etc., and can be used for scene applications based on the fused sensed information.

[0076] In one embodiment, see Fig.16 , this application also provides a smart intersection system network architecture for a smart intersection. The details are as follows: This smart intersection system involves data transmission of intersection collection equipment, electronic police, signal control smart intersection system, etc., and the security of the transmission line needs to be considered. According to relevant requirements, a suitable smart intersection system network architecture should be designed and equipped with corresponding network security equipment. There are two main transmission methods for new equipment at the intersection. One is wired transmission based on the existing signal networking network of the Traffic Management Bureau (the laser radar and video detector data are transmitted back to the signal timing center of the Traffic Management Bureau through the existing signal networking network switch and signal machine data at the intersection, and the signal timing center and the intersection signal machine realize two-way wired transmission). The other is wireless transmission (the laser radar and video detector are directly wirelessly transmitted back to the signal timing center of the Traffic Management Bureau). During construction, wired transmission is used for intersections that meet the wire-threading conditions; for intersections where the pipeline does not meet the requirements or the road excavation cannot be implemented or the road excavation cost is too high, wireless transmission is used. Wireless transmission must be based on the existing signal networking network of the Traffic Management Bureau, based on the 4GTD-LTE international standard, providing high-speed data services, with rich security services, and providing timely protection to defend against potential security risks and illegal intrusions to ensure high security.

[0077] In this architecture, the access switch is responsible for the access of various roadside facilities within a single point (pole), and accesses RSU, cameras, millimeter-wave radars, lidars, variable information boards, weather stations, etc. through the access switch. Multiple access switches and an aggregation switch (or aggregation router) form an access fiber ring network, and the roadside edge computing equipment is connected to the aggregation switch (or aggregation router) on the ring network. The ring network should be a GE ring. If the traffic demand is large, it can be expanded to a dual GE ring. When it is unable to form a fiber ring network, consider using a chain or star network. The aggregation switch (or aggregation router) can be connected to the core switch (or core router) in a dual-regulation or ring network.

[0078] In one of the embodiments, the smart base station in this application is divided into two categories according to its core components, namely, multi-source data fusion detection using multi-line radar combined with video information and multi-source data fusion detection using millimeter-wave radar combined with video information. Specifically, the smart base station (laser) consists of a 32-line laser radar, 3 cameras, 1 V2XRSU device and an edge computing node, and the smart base station (millimeter wave) consists of a millimeter-wave radar, 1 camera, 1 V2XRSU device and an edge computing node. On this basis, the corresponding network bandwidth resources need to be matched.

[0079] Table 1 Bandwidth of smart base station equipment

[0080] Furthermore, considering the bandwidth load and giving a certain redundancy value (30% of the total bandwidth), the minimum bandwidth requirement of a single smart base station (laser) can be calculated as: (3*4+20+2+5)*1.3=50.7Mbps. Similarly, the minimum bandwidth requirement of a smart base station (millimeter wave) is: (4+5+2+5)*1.3=20.8Mbps.

[0081] At the same time, the storage capacity requirements of the smart intersection system also need to be considered (the hard disk requirement is based on 180 days and 20% redundancy as an example, and the configuration of storage resources is considered. If there are other specific requirements, corresponding adjustments will be made).

[0082] Table 2 Data storage capacity of smart base stations

[0083] According to the bandwidth information of the smart base station, it can be calculated that the data volume per second of a single smart base station (laser) is: 50.7 / 8=6.3375MB, and the daily storage data is 6.3375*3600*24 / 1024=534.7GB / day. Taking into account the future business development, 20% of data storage redundant space is reserved. To meet the storage needs for at least 90 days, the required storage capacity is at least 534.7*90 / 1024*1.2≈56.4TB; to meet the storage needs for at least 180 days, the required storage capacity is at least 534.7*180 / 1024*1.2≈112.8TB.

[0084] Similarly, it can be calculated that the data storage speed of a single smart base station (millimeter wave) is: 20.8 / 8=2.6MB / s, and the daily storage data is 2.6*3600*24 / 1024=219.4GB / day. To meet the storage needs for at least 90 days, the required storage capacity of the hard disk is at least 219.4*90 / 1024*1.2≈23.1TB; to meet the storage needs for at least 180 days, the required storage capacity of the hard disk is at least 219.4*180 / 1024*1.2≈46.3TB.

[0085] It should be noted that if off-site disaster recovery is required, the storage capacity requirement should be multiplied by 2.

[0086] In one of the embodiments, the present application also provides a method for making a high-precision map of an intersection, which realizes the construction of a centimeter-level high-precision map. The method uses a high-precision map to realize a spatial reference, and realizes the calibration of the time line through the timing of GNSS to form a space-time reference space, and realizes a space-time reference consistent with the space-time coordinates of the actual toll station, providing basic support for the digital reconstruction of smart intersections. The method includes: high-precision map acquisition, map data update, high-precision map mapping and three-dimensional modeling.

[0087] Among them, high-precision map collection includes: using high-precision map collection equipment to collect high-precision map data. The collection vehicle includes on-board laser radar, IMU, high-definition panoramic camera and other equipment, integrated with sensors such as wheel speed meter, equipped with professional collection tool platform, to carry out point cloud and real scene 3D map of road surface and attached objects, and after removing traffic participants, form real scene 3D data of road surface, including point cloud map and tile map. High-precision map data collection content: lane lines, lane center lines, road edge lines, traffic signs, traffic signal control equipment, monitoring equipment, guardrails, street lights, gantries and other road elements, as well as the main attachments around the road, including roadside green belts, trees, buildings and other key facilities spatial data.

[0088] Map data updating includes: updating the real-life three-dimensional data of the road surface and the vector data of the attachments.

[0089] High-precision map making includes: data preprocessing, point cloud extraction, attribute editing, data compilation and encryption, and data quality inspection. Data preprocessing: Through GPS data adjustment and solution, the collected point cloud information and image information are integrated with high-precision positioning information to generate preprocessed high-precision map data. Use aerial triangulation and other processes to preprocess real-life three-dimensional data. Point cloud extraction includes extracting point clouds of road feature information from preprocessed data through the high-precision map making software smart intersection system to achieve point cloud classification and image deduplication correction. Attribute editing includes assigning attributes to classified map data and producing vectorized map data. Data compilation and encryption is to compile vectorized map data into a data format that can be recognized by the platform and vehicles, and perform data encryption processing. Data quality inspection includes consistency checking and quality inspection processing on the compiled data.

[0090] 3D modeling is achieved through environmental modeling, traffic participant modeling, 3D rendering, high-precision map data processing, map fusion, scene layout, and weather smart intersection system special effects rendering. Among them, environmental modeling is based on high-precision map data to model static objects such as roads, buildings, traffic lights, cabinets, roadside V2X equipment, roadside cameras, roadside radars, signs, street lights, gantries, roadside poles, cones, water barriers, guardrails, LED screens, trees, etc. in real scenes. Traffic participant modeling includes making target vehicles (including but not limited to M1, M2, M3, N1, N2, N3, and O-type motor vehicles), pedestrians (including but not limited to adults and children), and non-motor vehicles (including but not limited to human-powered bicycles and electric bicycles) to realize dynamic traffic scene display and call. 3D rendering is to build a 3D scene based on high-precision maps and 3D modeling data, and support vehicle following and perception result rendering. High-precision map data processing includes the fusion of high-precision data and 3D models during 3D rendering based on road surface data, roadside scenery, and building location information provided by high-precision maps. Map fusion includes merging the original map data, including the processing of Tiles files, the construction of road meshes and the modification of road boundaries, and finally forming a complete map. Scene layout includes high-precision maps, which are image maps and height maps using the CGCS2000 coordinate system. Scene layout is based on the image data obtained during map acquisition to construct terrain scenes, road maps, real-life buildings and scene vegetation. The weather smart intersection system special effects rendering is divided into four types of weather (sunny, cloudy, rainy and foggy). The current weather conditions and time are automatically obtained through physical time, or the weather conditions and time are manually set to render the map effects.

[0091] The smart intersection system includes a digital middle platform, which integrates the edge computing processing results of the front-end global perception smart intersection system, virtualizes the real-time perception data of the intersection, and maps the operating equipment and traffic participants to a virtualized digital scene consistent with the real scene. On this basis, it realizes digital support for the business application scenarios of smart intersections and has the following functions: Data base: The data base is used for receiving, collecting, exchanging and statistically analyzing various types of data, including data access and management, data storage, data exchange and management, and data statistical analysis. Among them, data access and management include realizing fast access to intelligent road perception data, supporting access support for multiple protocols including but not limited to MQTT, HTTP, LWM2M, etc., which can meet both the real-time requirements of long connections and the data access requirements of short connections. Data storage includes basic static data and dynamic business data, including spatiotemporal database, dynamic traffic database, AI video database, laser point cloud database, environmental status database, video perception database, infrastructure database, and shared database. The spatiotemporal database provides time synchronization services for devices, the dynamic traffic database is the result data perceived by video sensors and smart base stations, the AI ​​video database and laser point cloud database are the repositories of raw data, retaining the raw data to provide support for subsequent event restoration and event forensics, the environmental database is the basic data for visual display of static map and environmental modeling data, and the shared database provides a storage environment for external databases for access to other data. Data exchange includes multiple types of data interactions, such as smart base station information interaction, traffic equipment interactive control, and induction equipment interaction. Among them, the information interaction of smart base stations means that the digital middle station can interact with the roadside smart base stations in real time, and issue instructions and information, including issuing traffic warning services, life information services, high-precision map services and high-precision positioning services. Edit the service information on the digital middle station, select the designated smart base station to issue, and display it on the roadside variable information sign, including the name of the smart base station, service type and service content. Traffic equipment interaction means that the digital middle station can interact with the roadside traffic perception equipment in real time, and the vehicle detector, traffic event detector and other equipment, through the specific point of the equipment and GNSS timing, make the detector and the digital middle station have a unified space and time reference, and transmit the traffic perception information (including traffic volume, average speed, occupancy, traffic events, etc.) to the digital middle station in real time. The digital middle station twins the traffic perception information in real time in the three-dimensional high-precision map. Control and induction equipment interaction means that the digital middle station can interact with the intersection control and induction equipment in real time, and issue instructions and information, including prompt information, speed limit information, induction information, warning information, etc. Edit the control and induction instructions on the digital middle platform, and select the designated control and induction device to issue, including the control and induction device name, number, warning / induction instruction type and instruction content, etc.

[0092] Equipment status management: The equipment status management module includes intersection equipment configuration management, equipment status monitoring, equipment failure alarm, etc. Equipment configuration management manages the configuration information of intersection electromechanical equipment (such as equipment type, number, name, location, parameter configuration, etc.). Equipment status monitoring monitors the status of intersection electromechanical equipment and issues an alarm when equipment fails (communication failure, equipment failure, component failure, etc.).

[0093] Digital twin: Combining high-precision maps with three-dimensional engines, based on a unified spatial and temporal reference, the real-time traffic scene and event information of toll booths are digitally reconstructed on the digital platform in real time, providing a digital scene consistent with the real scene. The smart intersection system can realize point twins and global twins. Among them, point twins are achieved by calibrating the absolute position of the smart base station and GNSS timing calibration, and initializing the smart base station using high-precision maps, so that the detection results of the smart base station (position information, heading information, speed information, size system information, etc.) have space and time synchronized with the digital platform. At the same time, the smart base station transmits the perception results (point cloud data, real-time video data, recognition result data) to the digital platform in real time, and the digital platform twins the recognition results (such as real vehicles, pedestrians, etc.) in real time on the three-dimensional model of the high-precision map, realizing the synchronous display of point cloud data and video data. Global twinning is achieved by calibrating the absolute longitude and latitude of the smart base station, calibrating the GPS timing, and initializing the base station using a high-precision map to ensure that the base station detection results (location information, heading information, speed information, size system information, etc.) have spatial and temporal information synchronized with the platform. At the same time, the smart base station transmits the perception results to the platform in real time. The platform receives roadside base station data as: point cloud data, real-time video data, and recognition result data. The platform twins the recognition results such as real vehicles and pedestrians in real time on the three-dimensional model of the high-precision map, and displays the point cloud data and video data on the platform synchronously. Data statistics and analysis: Multi-dimensional statistics and analysis of various types of data are carried out to provide the content required for upper-level scenario applications on the one hand, and to provide managers with relevant decision-making auxiliary analysis on the other hand.

[0094] Automatic data upload: The smart intersection system has the function of automatic data upload and retransmission. The edge computing unit has a built-in hard disk and adopts an automatic cyclic overwriting data storage mechanism. When the storage reaches the maximum storage capacity, it will automatically perform a cyclic overwriting. When the dedicated line network failure between the smart base station and the central management cloud platform causes data transmission to be interrupted, the smart intersection system will continue to temporarily store data in the storage medium and automatically resume transmission back to the back-end management cloud platform after the network is restored.

[0095] Remote configuration and maintenance: The smart intersection system supports on-site configuration through the human-computer interaction interface and remote configuration at the center. It can set parameters such as detection area, alarm threshold, data upload time interval, etc.; remotely and automatically query terminal information (terminal model; software version number), and remotely upgrade the software; remotely obtain the device status of the smart base station; and realize remote maintenance such as remote restart and reset of the smart intersection system.

[0096] In one of the embodiments, a method for controlling an intersection is also provided, which can be implemented through an integrated control platform. The basis for the implementation of this method is holographic situational awareness of the intersection area and holographic perception of the intersection (digitalization of the intersection). For specific implementation, please refer to the above. The above provides a reference solution for how to realize the equipment layout of the digital intersection, the smart intersection system architecture, and data processing. The same or similar content will not be repeated here.

[0097] This control method achieves real-time and accurate perception of traffic intensity, commuting time, traffic mode share, road network status, etc. within the intersection by reasonably deploying multi-source fusion perception, edge computing and other equipment at intersections, combined with cutting-edge technologies such as artificial intelligence, big data, cloud computing, and digital twins. It accurately depicts the time domain evolution of intersection traffic, provides real-time analysis, judgment, prediction, optimization and other intersection control decision-making references for refined traffic management, and accelerates the digital transformation of traffic intersections.

[0098] The smart intersection system is based on centimeter-level high-precision maps and establishes a three-dimensional model of real scenes. It combines multi-source fusion data perception and edge computing to analyze the "full-time, multi-dimensional, and full-factor" of "people, vehicles, roads, environment, and events" at the intersection. A digital twin smart intersection that is consistent with the current time and space environment of the intersection is built in the center, which can accurately depict the driving trajectories of motor vehicles, non-motor vehicles, and pedestrians, and present real-time dynamic presentation and efficient intelligent analysis of core businesses such as intersection traffic flow characteristics, traffic signals, traffic hazards, traffic incidents, and traffic conflicts, thereby realizing the refinement, intelligence, and standardization of traffic intersection management.

[0099] On the one hand, the smart intersection system can perceive in real time macro-indicator information of intersections, such as traffic volume, instantaneous speed, average speed, time occupancy, headway and vehicle spacing; on the other hand, it can collect real-time micro-precise parameters such as the type of traffic participants, precise location, driving speed, heading angle, driving trajectory, etc., and make a multi-dimensional holographic portrait of all people, vehicles and roads at the intersection, create an accurate and real-time digital base for the intersection, establish a dynamic digital intersection detection and optimization model, and improve the refinement and granularity of intersection control.

[0100] The smart intersection system obtains real-time intersection traffic status indicators through multi-source data fusion detection, and combines the preset AI analysis model to obtain the multi-dimensional operation status of the intersection in real time, including traffic flow, micro-characteristics of traffic participants, traffic operation status at the intersection, traffic violations and traffic accidents. Among them, the traffic flow characteristics are lane-level basic traffic data, including flow, saturation, occupancy, queue length, average speed, headway, number of stops, delays, etc. The micro-characteristics of traffic participants include the position, speed, heading angle, license plate number, position, outline, body color, posture, etc. of pedestrians, motor vehicles and non-motor vehicles. The traffic operation status of the intersection is the lane-level traffic flow, capacity, operation efficiency and abnormal operation status of the intersection, among which the abnormal operation status of the intersection includes overflow, congestion, deadlock, imbalance, blocking point / conflict point, empty degree, etc. Traffic violations describe the traffic order at intersections, including not only motor vehicle violations such as driving against traffic and running red lights, but also violations such as non-motor vehicles and pedestrians crossing the intersection, running red lights and crossing the line. On this basis, information can be displayed through the platform. After the information is displayed, it is convenient for management departments or personnel to manage. Based on the deployed roadside smart base stations, the smart intersection system can accurately detect traffic events in combination with high-precision maps and traffic regulations. The detected events may include: motor vehicle violations, such as slow driving, speeding, illegal parking, driving against traffic, illegal lane changes, illegal overtaking, illegal meeting, illegal rushing, illegal occupying of the road, driving without following signs and markings, and not giving way to pedestrians; non-motor vehicle violations, such as non-motor vehicle driving against traffic, driving on the wrong road, staying in the motor vehicle lane, running red lights, etc. Pedestrian violations, such as pedestrians crossing the road and running red lights. The comprehensive violation volume at the intersection presents the number of violations by motor vehicles, non-motor vehicles and pedestrians at the intersection according to different time periods, supporting the management department to comprehensively evaluate the safety indicators of the intersection.

[0101] The smart intersection system can monitor and record the location, time and process of three types of traffic accidents: motor vehicle accidents, non-motor vehicle accidents and pedestrian accidents. At the same time, the smart intersection system supports three-dimensional visualization of historical accidents on the platform side, providing data support for the traffic management department to determine the cause of accidents and prevent accidents. The statistical display content includes the number of motor vehicle accidents, non-motor vehicle accidents, pedestrian accidents, common accident locations, historical event restoration, etc. The smart intersection system integrates intersection laser radar, bayonet cameras, electronic police, edge computing and other perception and computing equipment to realize the perception of all factors such as road vehicles, pedestrians, environment, traffic events, etc., and performs multi-source data fusion based on edge computing nodes. The core traffic parameter detection accuracy is ≥98%, which greatly improves the reliability of the smart intersection system.

[0102] In one of the embodiments, a method for digital reconstruction of identity (traffic participants) is also provided, including: based on multi-source perception fusion technology, on the basis of high-precision maps, combined with edge computing nodes, the multi-dimensional information such as the identity, position, outline, trigger time, posture, etc. of traffic participants collected by sensing devices such as lidar, electronic police, and face capture cameras are synchronously integrated in time and space to characterize the digital identity characteristics of traffic participants.

[0103] Optionally, it also provides a service based on digital reconstruction of vehicle identity, which can realize global tracking of traffic participants at intersections based on identity information, event warning based on location information, and traffic behavior prediction, enabling digital management and service capabilities of traffic intersections. Optionally, the smart intersection system automatically identifies and saves the digital identities of all vehicles passing through the intersection, and centrally and effectively manages vehicle traffic information and traffic flow data, providing effective technical support for traffic safety warnings and traffic accident prevention. At the same time, the smart intersection system can provide the historical behavior trajectory of completed event vehicles, providing objective and accurate basis for the traffic police department to handle traffic incidents.

[0104] The smart intersection system can automatically record the type, location, speed, size, direction, movement trajectory and other information of traffic participants at the intersection, support GPS / Beidou functions, collect information in real time and report it to the management center. When a traffic violation is found at the intersection, it can record multiple violations. The smart intersection system supports violations such as driving in the wrong direction, changing lanes illegally, crossing the line illegally, driving in the wrong lane, and running red lights. At the same time, the central platform can restore the accident scene in three dimensions based on the historical vehicle behavior trajectory combined with video information, providing new law enforcement evidence support for the traffic management department. At the same time, the center can track and present the historical driving trajectory of a specified vehicle from the perspective of the vehicle based on the global vehicle positioning capability of the traffic intersection.

[0105] In one of the embodiments, the present application also provides a method for detecting traffic violation events, which includes: the smart intersection system uses the real-time vehicle location information collected by the multi-source fusion sensing device, combined with the high-precision map of the intersection and the digital twin intersection, to intelligently analyze and determine whether there are illegal U-turns, illegal right / left turns, driving in the opposite direction, pressing the line, crossing the line, violating prohibited lines and other illegal behaviors. At intersections where right / left turns are prohibited, illegal behaviors such as right / left turning vehicles and driving in the opposite direction can be tracked and determined, and the illegal vehicles can be photographed and their operation trajectory data can be collected for evidence.

[0106] Optionally, the violation event can be driving out of lane, which means that the vehicle does not drive in the specified lane when encountering a "directional driving lane", including left-turning or right-turning vehicles occupying the straight lane, or driving straight in the left-turn or right-turn lane, etc. The smart intersection system supports the recording of such violations, tracking and judging, and capturing and collecting operation trajectory data of the violating vehicles.

[0107] Optionally, the violation event can be an illegal lane change. If there is a solid white line or double yellow line between two parallel lanes at the intersection, the vehicle cannot cross the lane line and change positions between the two lanes at will. The smart intersection uses the high-precision map of the intersection and the real-time vehicle location information collected by the smart base station to track and judge, intelligently analyze and judge whether there is an illegal lane change, and record the illegal behavior of the vehicles involved, and capture the illegal process and retain the operation trajectory data for evidence.

[0108] Optionally, the violation may be a vehicle failing to give way to pedestrians; the smart intersection system can monitor and record vehicles failing to give way to pedestrians (at zebra crossings). In a complex intersection environment, it can track the trajectories of vehicles and pedestrians in the detection area, automatically analyze whether there are pedestrians passing through the area near the motor vehicle (the set area), and automatically record three pictures at different time locations to reflect the process of a motor vehicle failing to stop to give way to pedestrians, at least one of which has information on the license plate information that can be clearly observed.

[0109] Optionally, the violation event can be a pedestrian violation. The smart intersection system accurately locates pedestrians through the front-end smart base station and combines the signal light information to realize the violation types of pedestrians detected in the detection area, including: standing in the pedestrian detection area when the pedestrian crossing signal light is red, pedestrians running red lights, pedestrians not walking in the zebra crossing area, etc. The movement trajectory of related violation behaviors can provide a new violation evidence data for the traffic control department, and the roadside sound and light warning facilities can be used to broadcast warning and warning information in real time based on the location of the illegal pedestrians.

[0110] Optionally, the violation event can be driving on the lane line. The smart intersection system supports detection, capture, recording and identification of vehicles that drive on the lane line. It mainly targets vehicles that drive on the lane line for a continuous period of time. Such vehicles will affect the judgment of the drivers of the front and rear vehicles and interfere with the normal driving of the front and rear vehicles.

[0111] Optionally, the violation event can be driving in the opposite direction, which means that the vehicle's driving direction is opposite to the direction specified by the lane. The smart intersection system supports the detection, capture, recording and identification of illegal vehicles driving in the opposite direction. Prohibited driving, Optional, the violation event can be prohibited parking. The smart intersection system automatically records vehicles that violate the prohibited driving and prohibited parking rules by intelligently judging the direction of vehicle driving. For example, motor vehicles occupy non-motor vehicle lanes, illegally occupy bus lanes, and large vehicles occupy small car lanes. Some intersections will be set to prohibit left or right driving. For such intersections, vehicles that violate the traffic light instructions are considered illegal vehicles. The smart intersection system automatically records the illegal behavior of illegal left-moving motor vehicles by intelligently judging the direction of vehicle driving, and saves and uploads the movement trajectory of the illegal vehicle to the central platform.

[0112] The smart intersection system can use the front-end roadside lidar to achieve accurate contour information of vehicles within the detection range, and at the same time combine the license plate color and video detection data fusion technology to determine the vehicle type, color, purpose, etc.

[0113] The smart intersection system can upload the traffic data collected by the smart intersection system to the detachment command center for archiving and application in real time according to user settings; the violation data is sent to the traffic police department's violation database in real time. The violation data includes: the time, location, lane, movement trajectory (laser point cloud) of the vehicle passing, full picture of the vehicle and road environment (2 to 3 pictures), close-up picture of the license plate, violation type, license plate number (automatically recognized by computer), license plate color, etc.

[0114] At the intersection, the front-end edge computing device uploads the result data and original data to the command center database in real time after data fusion and preliminary judgment of the violation, speeding up the processing speed and processing capacity of the smart intersection system for violation punishment.

[0115] The smart intersection system has the function of resuming data transmission from breakpoints: when a network failure occurs, the smart intersection system can save data, pictures, etc. at the front end of the intersection. After the failure is eliminated, the smart intersection system will automatically upload the data, pictures, etc. to the traffic police data center.

[0116] The smart intersection system provided by this application accurately senses road conditions through roadside smart base stations. The smart intersection system records pedestrian crossing information and oncoming vehicle information at intersections, and supports high-precision location information of traffic participants. At the same time, the type, location, speed, size, direction, movement trajectory and other information of traffic participants are recorded in real time and uploaded to the management center. When a traffic accident occurs at an intersection, the three-dimensional restoration of the scenes before, during and after the accident is achieved by combining the vehicle's running position trajectory and the original video data. At the same time, the platform can track the driving trajectory of the specified vehicle, realize automatic tracking of key vehicles, and provide new law enforcement evidence collection methods for traffic management departments.

[0117] The smart intersection system provided by this application detects, captures, recognizes license plates, and records passing vehicles in the monitored lanes during all phases of the monitored direction, as well as performs traffic flow statistics and other functions. The basic traffic parameters such as cross-sectional traffic flow are obtained through the vehicle detection module and written into the central database for easy query and statistics, providing auxiliary decision-making for traffic managers.

[0118] The smart intersection system provided in this application can perform statistical analysis on pedestrian traffic violations / illegal incidents at a certain intersection, and display the changes in the number of traffic violations / illegal incidents within a day in the form of visual charts.

[0119] The smart intersection system provided by this application can process and analyze the traffic flow data collected by the smart base station at the intersection in real time, and use big data technology to integrate and calculate the traffic flow data of different time granularities and different spatial granularities, obtain the intersection traffic operation index, and obtain the road traffic status. Compare the smart intersection in multiple time periods, including the traffic conditions throughout the day, the traffic conditions during peak hours in the morning and evening, or the traffic conditions overview and congestion index changes during other time periods. Provide a comparison of data such as congestion index, congestion level, average speed of road sections, length of congested road sections (lanes), and congested mileage ratio, and provide a comparison of traffic index line charts for different time periods throughout the day. For special events and special periods of large-scale activities, by conducting real-time or dynamic analysis and evaluation of key areas and road intersections in different time periods such as before, during, and after the event, managers and decision makers can grasp the objective status quo, changing trends, and laws of traffic operation, thereby providing an intelligent, automated, and efficient analysis platform and data report for the formulation of traffic security policies and transportation service plans.

[0120] The smart intersection system provided by this application can analyze the traffic flow gathering rules at intersections. Based on the traffic collection data, the traffic flow change rules, traffic flow similarity and key traffic flow identification can be analyzed respectively. The traffic flow change rules require the analysis of the inflow and outflow flow of the intersection by year, month, day, hour, 30 minutes, 15 minutes, 10 minutes and 5 minutes. The traffic flow similarity is based on 5 minutes as the basic event granularity.

[0121] The smart intersection system provided by this application can realize the analysis and judgment of traffic safety at intersections. Specifically, it can realize: analysis and judgment of the spatiotemporal characteristics of traffic accidents, analysis and judgment of the composition of traffic accidents, analysis and judgment of the causes of traffic accidents, analysis and judgment of the situation of traffic accidents, and analysis and judgment of traffic accident black spots. The analysis and judgment of the spatiotemporal characteristics of traffic accidents is to analyze the location and time period (daytime, nighttime, morning peak, evening peak, flat peak, etc.) of road traffic accidents. At the same time, the high-precision map of the smart intersection system is associated to display the spatial distribution of traffic accidents in different time periods in the digital intersection. The analysis and judgment of the composition of traffic accidents is based on the precise vehicle positioning information and movement trajectory collected by the roadside smart base station, and through the preset traffic accident judgment algorithm, the type of road traffic accident, the type of vehicle involved, the movement trajectory of the accident vehicle before the accident, the objects of the accident parties (vehicle-to-vehicle, vehicle and pedestrian, motor vehicle and non-motor vehicle, vehicle accident itself), accident impact area and other data are analyzed. The analysis and judgment of the causes of traffic accidents is to realize the analysis of the causes of road traffic accidents based on the front-end multi-source sensing equipment being able to accurately locate traffic participants, combined with high-precision maps of intersections and time-space positioning technology, such as analyzing frequent lane changes, driving without following traffic lights, and driving in the wrong direction, to provide data support for road traffic safety publicity and traffic accident prevention. The analysis and judgment of traffic accident situation is to conduct a month-on-month and year-on-year analysis of traffic accidents in a certain area over a certain period of time (week, month, quarter, year, etc.), to provide data support for the analysis and judgment of traffic safety situation, and can also be used as a basis for assessing the road traffic safety management work of the brigade and squadron in the jurisdiction. The analysis and judgment of traffic accident black spots is to analyze the number, type and cause of road traffic accidents at a certain intersection or section in the past period of time (week, month, quarter, year, etc.), to provide data support for the investigation of accident black spots and accident prevention work.

[0122] The smart intersection system proposed in this application can analyze the evolution patterns of traffic conditions, including intersection congestion periods, congestion duration, intersection operation status index analysis, and frequent congestion point analysis.

[0123] Traffic congestion analysis includes: associating traffic flow data with intersection lane type data, and analyzing the influence of intersection lane type on road congestion based on the intersection lane type data when the road section is unobstructed and congested. The smart intersection system accurately perceives the road conditions based on the roadside smart base station and provides real-time traffic information, such as lane traffic volume, occupancy rate, queue length, etc. By selecting different indicators (such as intersection congestion rate, average vehicle speed, and vehicle queue length, etc.) to analyze the operation status of the intersection, from a time perspective, form early warning indicators for whether the intersection is congested, and determine the critical point of congestion formation and dissipation. Analyze the road congestion situation, analyze the intersections that are often congested, and analyze the high-incidence periods of congestion, to provide a strong data basis for congestion prediction and relief.

[0124] Real-time congestion analysis and judgment is to calculate the average vehicle speed, congestion duration, congestion distance and other data based on the traffic congestion situation of urban roads, and present the traffic congestion situation in real time on the digital twin intersection display page of the platform in the form of a heat map. At the same time, the real-time congestion situation is published to the traffic induction device. According to the set threshold, the congestion situation is alarmed according to three levels: severe congestion, moderate congestion and general congestion, and the congestion warning information is pushed through the platform interface of the relevant management department to take diversion measures in time.

[0125] Traffic flow prediction and analysis services are used to predict and analyze road congestion through in-depth mining, analysis and judgment of historical traffic flow data. They can predict road traffic congestion at a specified time (in hours). They can predict congestion at a specified time range on a specified road and analyze the impact trend of congested intersections within a specified time.

[0126] The smart intersection system of the present application is connected to a traffic signal controller that supports intelligent networking functions, and obtains the operating status of intersection signals in real time (including signal cycle duration, phase sequence design, green-to-signal ratio allocation, etc.). Combined with the smart base station, it detects accurate data on traffic flows in all directions of road intersections (such as import lane speed, congestion time, queue length, etc.), providing data support for the optimization of traffic signal control strategies to achieve the goal of balancing traffic pressure.

[0127] The smart intersection system of the present application realizes all-weather intersection imbalance monitoring and updates at a frequency of ≤2min. The so-called intersection imbalance refers to the situation where the saturation of each entrance road of the intersection is greatly different due to the uneven distribution of traffic demand in a certain period of time, and even the situation where traffic in a certain import direction is saturated while traffic in other directions is not saturated. Saturation is the ratio of traffic flow to capacity. The smart intersection system can calculate it based on the real-time traffic flow data obtained and the current traffic signal control scheme. The intersection imbalance index refers to the maximum saturation value of the intersection. Intersection imbalance often occurs with morning and evening rush hours, and the frequency of occurrence is higher at intersections in sections with tidal phenomena; the frequency of occurrence is also high in some large commercial areas and near large sports facilities where events are held. Intersection imbalance will not only greatly reduce traffic efficiency but also may cause serious traffic accidents. Therefore, it is of great significance to monitor intersection imbalance in real time around the clock. The smart intersection system calculates and analyzes whether the intersection is unbalanced, cooperates with traffic signal control smart intersection system, and information induces smart intersection system to achieve the goal of balancing traffic pressure.

[0128] The smart intersection system of this application can realize all-weather intersection overflow monitoring at intersections throughout the city, and update it at a frequency of ≤2min. Intersection overflow refers to the situation in which, within a certain period of time, due to excessive traffic demand at the intersection, the green light time of a certain import direction is insufficient, and vehicles cannot be fully released, resulting in the arrival of vehicles parking twice or multiple times and queuing. After several cycles, the queued vehicles in the import direction overflow to the upstream intersection. The intersection overflow index is the number of vehicles overflowing to the upstream intersection, which can be obtained by obtaining the current queue data and the maximum non-overflow queue data recorded by the smart intersection system through real-time monitoring of traffic signals. Intersection overflow often occurs near commercial areas, commercial centers, hospitals, and schools during school hours with huge traffic flow. Some intersections where traffic accidents have occurred will also cause intersection overflow. Intersection overflow greatly affects the traffic efficiency of downstream intersections, and then spreads the overflow to other intersections, forming a chain deadlock, large-scale congestion, and may also cause traffic accidents, threatening the lives and safety of traffic travelers. Therefore, real-time monitoring of intersection overflow is very necessary. The smart intersection system determines whether the intersection is overflowing through calculations, controls the smart intersection system with traffic signals, and uses information to guide the smart intersection system to achieve the goal of balancing traffic pressure.

[0129] The smart intersection system of this application can realize the single-point operation evaluation of traffic signals at intersections. The evaluation indicators are scored by establishing reference values ​​for each evaluation indicator. The indicators of the evaluation model include the running speed of vehicles passing through the intersection, the average number of stops, the average queue length, the average waiting time, and the intersection saturation, etc. Each evaluation indicator is obtained through data analysis, and compared with the reference value of each evaluation indicator to obtain the score of the indicator. By comprehensively analyzing the scores of various indicators, an overall evaluation of the operation of the traffic signal at the intersection is obtained.

[0130] There are advantages and disadvantages to installing and operating traffic signals to control conflicting traffic and pedestrian movement at intersections. Advantages include the potential reduction of certain types of collisions (particularly corner collisions), provision for pedestrian crossings, provision of roadside vehicles into the traffic flow, possible capacity improvements, and possible delay reductions. However, signals are by no means a perfect solution to the problem of delays or accidents at an intersection. A poorly timed signal or an irrational signal The display of the various signal indications (green, yellow, red, protected, permitted) at an intersection is handled by a signal controller (usually located in a cabinet next to the intersection). Modern signal controllers are sophisticated electronic devices. These controllers, when combined with methods of vehicle detection, provide a great deal of flexibility in controlling phase durations and sequences. Traffic signal controllers are designed to operate in one or more of the following modes: scheduled, semi-real time, or real time.

[0131] Scheduled Time: A signal whose timing (cycle length, green time, etc.) is fixed for a specified period of time and does not change as traffic flow at the intersection changes. This mode of operation does not require vehicle detection.

[0132] Semi-real-time: A signal whose timing (cycle length, green time, etc.) is affected when a vehicle is detected (via video, pavement embedded inductive loop detectors, etc.). On some methods, but not all. This mode of operation is often seen where low-capacity roads intersect with high-capacity roads, often referred to as minor and major streets, respectively. In this case, green time is assigned to the major street until a vehicle is detected on the minor street; then the green indication is briefly assigned to the minor street before returning to the major street.

[0133] Real Time: A signal method where timing (cycle length, green time, etc.) is completely affected by traffic volume. Most commonly used at intersections with two major streets and all approaching traffic volumes that vary widely throughout the day.

[0134] Most existing intersection signal control schemes use a preset timing scheme. While preset timing signal control does not require the expense of vehicle detection hardware, it results in signal timing that is not responsive to real-time traffic demand. The fixed time values ​​for preset timing signals are based on expected traffic demand during the time period of interest. However, traffic flows vary greatly from cycle to cycle. Therefore, with fixed timing, one phase may provide too much green time (which will cause additional delays for vehicles moving in other phases) and not enough time in another cycle. Over the past 20 years, improvements in signal controller technology have made it possible to make significant progress in traffic control at signalized intersections. Modern signal controllers are able to accept inputs of traffic demand and use this information to adjust the green interval duration and phase sequence from one cycle to the next.

[0135] Real-time control requires detection technology, and for motor vehicles, the most common form of vehicle detection technology is the Inductive Loop Detector (ILD). ILD is a simple technology that has been used for decades and consists of a loop (or coil) of wire embedded in the road surface, through which an electric current circulates. This current is monitored by a device that is interfaced to the signal controller, and when a vehicle passes through the ILD, the inductance level of the current is changed. When the monitoring device detects a change in the inductance level, it sends an input to the signal controller to indicate the presence of a vehicle. The limitation of ILD technology is that vehicles can only be detected at the location of the ILD. This limits control options because it is costly to lay a large number of ILDs at intersections.

[0136] The smart intersection system of the present application accurately senses the road conditions and roadside traffic facilities through roadside smart base stations deployed at urban road intersections, and controls road traffic signals based on lane-level historical data and / or real-time data in combination with the ever-changing traffic conditions at urban road intersections (Note: a traffic signal controller with support for intelligent networking functions), thereby enabling special vehicles (such as fire trucks, ambulances, etc.) and public transportation vehicles to have priority passage.

[0137] In one of the embodiments, the present application provides a method for adaptive optimization of intersection signal control, the method comprising: the intelligent intersection system uses detection data to predict macro traffic patterns, and uses the predicted data to optimize the cycle and green light time of the intersection. The green light time is optimized again according to the actual arrival of the vehicle (sensing principle). The multi-source fusion perception equipment at the roadside intersection can perceive the traffic flow in all directions of the intersection in real time, and then send the perception results to the networked signal machine through the roadside RSU. The signal machine automatically increases or decreases the green light timing according to the real-time traffic flow and the road right priority division, thereby reducing the green loss time at the intersection and improving the traffic efficiency at the intersection.

[0138] In one embodiment, see Fig.17 , the present application provides a coordinated bus priority signal light control method. Bus priority control is to ensure that buses have priority while ensuring the coordination effect of the main road. This method detects the bus through the roadside sensing smart intersection system and predicts the time when it arrives at the intersection stop line. According to the predicted signal light status of the intersection when arriving, the signal light is optimized and controlled in advance without affecting the green wave of the main line, so that the bus can pass through the intersection without stopping, and the rapid passage of public transportation is guaranteed to the greatest extent. The bus priority control method provides methods such as extending the green light and turning on the green light in advance. Accordingly, in one of the embodiments, the present application provides a traffic light control method with special service priority. Through the special service priority control function, the smart intersection system can support special vehicles (such as fire trucks, ambulances, etc.) and public transportation vehicles. Based on the digital identity information of the special vehicles, the traffic status of the road ahead can be perceived in advance. The edge computing unit quickly analyzes and calculates the priority traffic strategy, and deploys the strategy locally to the signal control device through information release facilities (including V2X, sound and light warning equipment, information boards, etc.). At the same time, avoidance information is sent to vehicles in the affected area, so that the service vehicles can pass through the scheduled route with minimal delay along the scheduled route, thereby ensuring that the police convoy arrives at the destination on time and safely, and at the same time minimizing the impact on social vehicles.

[0139] In one embodiment, the present application provides a traffic incident warning method. When abnormal conditions at intersections are obtained through roadside smart base stations (such as large trucks, vehicles speeding too fast, running red lights, driving in the wrong direction, etc.), warning information is issued to pedestrians through roadside alarm devices, and the information is delivered to the corresponding vehicles through V2X or other appropriate methods, prompting the driver to slow down or take risk avoidance measures. Optionally, the corresponding information can be uploaded to the central cloud platform.

[0140] In one of the embodiments, the present application provides a real-time intersection signal control solution for slow-moving traffic participant detection in response to the specific environment and requirements of the intersection.

[0141] The detection capability of the roadside smart base station provides a good and reliable data source and support for the realization of this real-time signal timing solution. In addition to completing the ILD function, it can also complete practical functions such as non-motor vehicle and pedestrian detection, queue length detection, etc., providing more dimensional data support for signal control. At some traffic intersections that mainly serve slow-moving traffic participants such as non-motor vehicles and pedestrians, the guarantee of this detection capability is indispensable and crucial.

[0142] For real-time solutions, due to the previous road detection capabilities, only the traffic volume of motor vehicles can be detected, so that signal timing mostly only considers the traffic efficiency of motor vehicles and ignores slow-moving traffic participants such as non-motor vehicles and pedestrians. This embodiment mainly considers the signal timing solution designed for motor vehicles and slow-moving traffic participants. The method includes: Obtain various parameters such as real-time traffic flow, queue length, non-motor vehicle and pedestrian flow of each lane at the intersection; analyze the flow rate ratio of each lane group according to the real-time traffic flow, determine the phase combination and the key traffic flow of each phase; calculate the optimal cycle length of the next cycle; calculate the flow rate ratio of the key lane group of each phase to allocate the green light duration for the next cycle; compare the calculated green light duration with the maximum and minimum green light durations, pedestrian protection time, etc., and take the appropriate value; Among them, the maximum green light duration means that even if the arrival rate of vehicles is high enough to extend the deadline indefinitely, the phase will eventually be terminated to serve other traffic flows. The initial green light time provides a practical minimum green time for traffic movement. The initial green time is intended to serve all or most of the vehicle queues developed during the red period of the traffic movement. Vehicles that start queuing at the green start, or vehicles that arrive after the queue is cleared, are usually processed during the extension period. Therefore, a green time that is large enough should be specified to provide a regularly expected initial queue length, but not too large to avoid wasting too much green time in cycles with far lower average vehicle arrivals. In addition, the green time should be at least a few seconds more than the loss time of a stage. The typical minimum green time is about 10 seconds.

[0143] For different traffic light control objectives, different solutions can be designed based on indicators such as queue length, arrival rate, average speed, etc. for different types of vehicles or pedestrians: Considering pedestrian crossing, when the phase changes, pedestrians are likely to be trapped in the middle of the intersection. To avoid this problem, the minimum green time required for pedestrian passage should be checked with the green time allocated for that phase. If there is not enough green time for pedestrians to cross the road safely, the allocated green time should be increased to meet the needs of pedestrians. If a pedestrian button (for drive control) is provided at the intersection, the green time can be increased to meet the needs of pedestrian passage only when the button is activated. The minimum time for pedestrian passage is calculated as follows:

[0144] in, Gp = the shortest time required for pedestrians to pass (s), 3.2 = Pedestrian preparation time to start passing (s), L = length of pedestrian crossing (ft), Sp = walking speed, usually 3.5 ft / s, Nped = number of pedestrians, WE = effective width of pedestrian crossing. It should be noted that, further considering that there are a large number of slower pedestrians (elderly, visually impaired, etc.) at the intersection, using a slower walking speed may be necessary.

[0145] Similarly, signal control for non-motorized vehicles can be considered. Of course, in actual situations, a more suitable signal control solution can be obtained by performing game calculations on pedestrians and non-motorized vehicles at the intersection based on the actual target detection of the roadside smart base station.

[0146] The signal control method provided in this application can reduce the total delay of the intersection, reduce the number of ILDs, reduce a lot of costs, and improve the efficiency of slow traffic participants. 2. Different real-time control effects can be achieved for different optimization goals. In one of the embodiments, the present application provides a method for warning of abnormal intersections, which includes a roadside smart base station accurately sensing data of traffic conditions at intersections, identifying the behaviors of various targets such as vehicles, pedestrians, and non-motor vehicles, and issuing timely warnings for abnormal conditions (such as the passage of large trucks, vehicles running red lights, vehicle failures, road construction, etc.), notifying non-motor vehicles and pedestrians at intersections through roadside information warning equipment, and also delivering information to corresponding vehicles through V2X or other appropriate methods, prompting drivers to slow down or take risk avoidance measures.

[0147] In one of the embodiments, the present application provides a method for blind spot / beyond visual range warning at an intersection. When a vehicle approaches the (traffic participant) blind spot (intersection of ultra-long visual range and non-visual range) section of an intersection, the roadside smart base station will provide accurate perception data to provide warning information (text, voice, warning lights) to pedestrians and non-motor vehicles preparing to pass through the intersection, and can also provide warning information to intelligent connected vehicles and autonomous driving vehicles.

[0148] In one of the embodiments, the smart intersection system supports multi-dimensional conditional query of historical data on intersection operation status, including lane traffic flow, average vehicle speed, congestion frequency, congestion duration, congestion spatial range, total congestion, etc. It can also monitor and analyze in real time, periodically conduct statistical evaluations on congestion points or unreasonable locations of road traffic, and report them to decision-making departments to identify weak links in road network operation, determine the root causes, and manage the urban traffic environment. It supports real-time operation status query, operation evaluation query, and historical operation status query.

[0149] In one of the embodiments, the smart intersection system supports query functions such as traffic flow, average lane speed, lane occupancy rate, lane queue length, turning rate, etc. of each lane at an intersection at any time, and supports the generation of statistical analysis reports such as traffic flow, speed, lane occupancy rate, lane queue length, etc.

[0150] In one of the embodiments, the platform side of the smart intersection system can display the traffic congestion status of the intersection in real time, and support comparison with the traffic operation status during the same period last week, and then evaluate the intersection situation. The evaluation indexes include congestion index, congestion mileage, average speed, city congestion index, etc., and can be presented in the form of tables.

[0151] In one embodiment, the smart intersection system can query violation data. The smart two-road intersection system of this application can sort and store violation data. It can conduct a comprehensive query on sorted, unsorted and manually entered violation data. In addition to displaying basic text information and evidence pictures of violation data, the query results will also prominently indicate data sorting status, sorter, sorting date and other data operation status information.

[0152] In one of the embodiments, the smart intersection system supports quick query of key traffic information such as historical intersection conditions, road operation index, average road speed, road traffic volume, queue length, road intersection congestion rate, (vehicle operation trajectory) in the form of a timeline.

[0153] In one of the embodiments, the present application provides a method for providing traffic condition prompts, including: accurately sensing road conditions through roadside smart base stations, and notifying pedestrians at intersections through roadside information warning equipment; connecting with vehicle-mounted / handheld APP data, and broadcasting in real time through V2X roadside equipment, so that intelligent connected vehicles and autonomous driving vehicles can receive the latest information on the road and traffic conditions ahead in real time, such as traffic accidents, road construction, traffic congestion, parking restrictions, and turning restrictions.

[0154] In one of the embodiments, the present application provides a method for managing traffic violations, including: real-time detection of motor vehicles and pedestrians at intersections through roadside smart base stations, capturing vehicles, pedestrians, and non-motor vehicles that have committed traffic violations. The smart intersection system integrates the traffic information, running direction, and quantity information of traffic participants to provide a new traffic management method for traffic management departments, thereby achieving management of motor vehicle violations and regulating drivers' driving behavior; at the same time, regulating the behavior of pedestrians and non-motor vehicles crossing the road at will, and building a harmonious traffic operation environment in which cities are civilized and law-abiding.

[0155] At present, the penetration rate of autonomous driving vehicles and connected vehicles is relatively low. The smart intersection system of this application can realize data sharing between roadside perception results and autonomous driving vehicles or connected vehicles. At the same time, considering the vehicle-road collaborative scenario, the smart intersection deploys edge computing units on the roadside, based on multi-source perception data fusion algorithms, AI intelligent analysis, deep learning and other technical means, to achieve perception supplement, safety warning, efficiency improvement prompts, information sharing and other services for autonomous driving vehicles, and promote the rapid implementation of vehicle-road collaborative autonomous driving scenarios.

[0156] In one of the embodiments, the smart intersection system of the present application can provide a perception supplement service for autonomous vehicles. The smart intersection deploys a holographic perception smart base station at the intersection, combined with a centimeter-level high-precision map digital twin platform, and has functions such as centimeter-level positioning, real-time perception of all elements of intersection information, and tracking of the movement trajectories of traffic participants, providing data support for traffic managers to manage intersections in a refined manner. At the same time, based on the V2X communication network, high-quality, low-latency data sharing between vehicles and vehicles, and vehicles and roads can be achieved. Since the autonomous driving vehicle's own perception range is limited, it cannot accurately perceive beyond the line of sight and has certain visual blind spots. By establishing a data sharing channel with the smart intersection, coordination between vehicles and roads can be achieved, providing autonomous driving with beyond the line of sight and blind spot supplementary perception capabilities.

[0157] In one of the embodiments, the smart intersection system of the present application can provide a safety warning method (collaborative decision-making), including: the smart intersection system integrates the information of traffic participants such as pedestrians and non-motor vehicles identified by the smart intersection system through front-end multi-source perception, judges the safety hazard information of traffic participants through multi-access edge computing deep learning, and interacts the warning information with the autonomous driving vehicle in real time, reminds the vehicle of the risk of collision, and recommends deceleration and avoidance to avoid the occurrence of traffic safety incidents.

[0158] Optionally, the safety warning service information that the smart intersection system in this application can provide is shown in the following table: Figure 1 Security warning service information

[0159] In one of the optional embodiments, the present application provides a traffic dispatch method at an intersection for improving the traffic efficiency at the intersection, including: based on holographic perception of traffic flow data, real-time analysis of traffic congestion and queue status at the intersection and information release through V2X roadside equipment, while combining dynamic lane management, traffic control information, traffic light messages, bus-only lane management and other information shared by networked signal machines, providing vehicles with speed guidance, dynamic lane adjustment, special service vehicle priority and other services based on the precise location of the vehicle, thereby improving the traffic efficiency of vehicles at intersections.

[0160] Optionally, the efficiency improvement service information that the smart intersection system of this application can provide is shown in the following table: Figure 2 Efficiency Improvement Service Information

[0161] In one of the optional embodiments, the present application provides a method for sharing intersection information, including: the smart intersection system constructs a digital twin model of the smart intersection to truly reproduce the microscopic information of the road intersection, including data such as signs and markings, traffic organization, channelization information, signal control strategies, intersection environment information, etc., and monitors the movement trajectories of traffic participants at the intersection in real time, and based on the precise position information of the autonomous driving vehicle, publishes information sharing services such as traffic signs at the intersection ahead, traffic prohibitions / restrictions, pedestrian courtesy, accident-prone spots, high-precision maps, differential data, etc.

[0162] Optionally, the information sharing service content provided by the smart intersection system of this application is shown in the following table: Figure 3 Information sharing service information

[0163] The smart intersection of this application can provide a "one-picture-all" traffic information service: the four major indicators of "intersection traffic situation", "intersection lane queue length", "intersection imbalance index" and "energy consumption intensity" can be viewed in real time through one picture to predict traffic conditions and aggregate and calculate carbon emissions at the intersection.

[0164] The smart intersection of this application can provide a "tracking all the way" traffic safety warning: it can monitor the behavior of traffic participants at the intersection in real time, perceive the identity information of the participants, trace the core problems of intersection congestion, and provide a key path for blind spot warnings at intersections, speeding warnings, abnormal intersection warnings, and traffic condition prompts for vehicles with networking functions.

[0165] The smart intersection of this application can provide "integrated intelligent management" intersection traffic control: through full collection of intersection data and full coverage of indicators, it can realize functions such as traffic signal coordination, violation behavior management, and traffic accident restoration, empowering the optimization of traffic light timing and traffic flow organization.

[0166] The smart intersection of this application can provide traffic management departments with refined intersection management methods, provide traffic participants with real-time intersection guidance services, enable autonomous driving vehicles to pass accurately based on roadside data, and drive the rapid implementation of autonomous driving technology in complex urban road environments.

[0167] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of one or more methods in the embodiments of the present application are implemented.

[0168] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of one or more methods in the embodiments of the present application are implemented.

[0169] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0170] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0171] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for environmental perception at an intersection, characterized in that: include: Acquire sensing data of multiple roadside smart base stations; wherein the roadside smart base stations are deployed at preset positions of intersections; each of the roadside smart base stations is aligned based on a high-precision map, and each of the roadside smart base stations is time synchronized based on a GNSS signal; Use registration parameters to spatially synchronize the sensing data of each roadside base station; The spatially synchronized perception data is calculated to obtain the environmental perception result of the intersection.

2. The environment perception method according to claim 1, characterized in that: The intersection is one of an intersection where a main road intersects another main road, an intersection where a main road intersects another secondary road, an intersection where a secondary road intersects another secondary road, and an X-shaped intersection, and the roadside intelligent base station is arranged diagonally.

3. The environment perception method according to claim 1, characterized in that: When the intersection is an X-shaped intersection, the roadside intelligent base station is deployed at the diagonal position where two roads are relatively closer to each other at the X-shaped intersection.

4. The environment perception method according to claim 1, characterized in that: The intersection is also equipped with roadside sound and light text warning equipment for outputting guidance information. The roadside sound and light text warning equipment is set based on the crosswalk on the road.

5. The environment perception method according to claim 1, characterized in that: The acquiring of the sensing data of multiple roadside intelligent base stations includes: Acquire lidar data and image data; Performing spatiotemporal synchronization on the laser radar data and the image data to obtain data to be fused; The data to be fused is fused and calculated to obtain the perception data of the roadside intelligent base station.

6. The environment perception method according to claim 1, characterized in that: The acquiring of the sensing data of multiple roadside intelligent base stations includes: Acquire millimeter wave radar data and image data; Performing spatiotemporal synchronization on the millimeter wave radar data and the image data to obtain data to be fused; The data to be fused is fused and calculated to obtain the perception data of the roadside intelligent base station.

7. The sensing method according to claim 5 or 6, characterized in that: The perception data includes the attributes, speed, heading angle and identity of the traffic participants in the intersection.

8. A smart intersection system, characterized in that: include: Multiple roadside smart base stations are used to obtain sensing data; wherein the roadside smart base stations are deployed at preset positions at intersections; each of the roadside smart base stations is aligned based on a high-precision map, and each of the roadside smart base stations is time synchronized based on a GNSS signal; The edge computing unit is used to spatially synchronize the perception data of each roadside base station using the alignment parameters; and calculate the spatially synchronized perception data to obtain the environmental perception result of the intersection.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Citation Information

Cited By

  • Transportation hub passenger flow control method based on holographic trajectory and related device

    CN119294768A

  • Intersection dynamic interaction information management method and system based on digital twinning

    CN121034102A

  • Intersection dynamic interaction information management method and system based on digital twinning

    CN121034102B

  • Multi-target continuous motion trail generation method and device for intersection scene

    CN121259046A

  • Intersection traffic simulation method and electronic equipment

    CN121705410A